CHANGES revision 194206
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 6 7 *) Don't set val to NULL when freeing up structures, it is freed up by 8 underlying code. If sizeof(void *) > sizeof(long) this can result in 9 zeroing past the valid field. (CVE-2009-0789) 10 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 11 12 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 13 checked correctly. This would allow some invalid signed attributes to 14 appear to verify correctly. (CVE-2009-0591) 15 [Ivan Nestlerode <inestlerode@us.ibm.com>] 16 17 *) Reject UniversalString and BMPString types with invalid lengths. This 18 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 19 a legal length. (CVE-2009-0590) 20 [Steve Henson] 21 22 *) Set S/MIME signing as the default purpose rather than setting it 23 unconditionally. This allows applications to override it at the store 24 level. 25 [Steve Henson] 26 27 *) Permit restricted recursion of ASN1 strings. This is needed in practice 28 to handle some structures. 29 [Steve Henson] 30 31 *) Improve efficiency of mem_gets: don't search whole buffer each time 32 for a '\n' 33 [Jeremy Shapiro <jnshapir@us.ibm.com>] 34 35 *) New -hex option for openssl rand. 36 [Matthieu Herrb] 37 38 *) Print out UTF8String and NumericString when parsing ASN1. 39 [Steve Henson] 40 41 *) Support NumericString type for name components. 42 [Steve Henson] 43 44 *) Allow CC in the environment to override the automatically chosen 45 compiler. Note that nothing is done to ensure flags work with the 46 chosen compiler. 47 [Ben Laurie] 48 49 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 50 51 *) Properly check EVP_VerifyFinal() and similar return values 52 (CVE-2008-5077). 53 [Ben Laurie, Bodo Moeller, Google Security Team] 54 55 *) Enable TLS extensions by default. 56 [Ben Laurie] 57 58 *) Allow the CHIL engine to be loaded, whether the application is 59 multithreaded or not. (This does not release the developer from the 60 obligation to set up the dynamic locking callbacks.) 61 [Sander Temme <sander@temme.net>] 62 63 *) Use correct exit code if there is an error in dgst command. 64 [Steve Henson; problem pointed out by Roland Dirlewanger] 65 66 *) Tweak Configure so that you need to say "experimental-jpake" to enable 67 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 68 [Bodo Moeller] 69 70 *) Add experimental JPAKE support, including demo authentication in 71 s_client and s_server. 72 [Ben Laurie] 73 74 *) Set the comparison function in v3_addr_canonize(). 75 [Rob Austein <sra@hactrn.net>] 76 77 *) Add support for XMPP STARTTLS in s_client. 78 [Philip Paeps <philip@freebsd.org>] 79 80 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 81 to ensure that even with this option, only ciphersuites in the 82 server's preference list will be accepted. (Note that the option 83 applies only when resuming a session, so the earlier behavior was 84 just about the algorithm choice for symmetric cryptography.) 85 [Bodo Moeller] 86 87 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 88 89 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 90 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 91 [Nagendra Modadugu] 92 93 *) The fix in 0.9.8c that supposedly got rid of unsafe 94 double-checked locking was incomplete for RSA blinding, 95 addressing just one layer of what turns out to have been 96 doubly unsafe triple-checked locking. 97 98 So now fix this for real by retiring the MONT_HELPER macro 99 in crypto/rsa/rsa_eay.c. 100 101 [Bodo Moeller; problem pointed out by Marius Schilder] 102 103 *) Various precautionary measures: 104 105 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 106 107 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 108 (NB: This would require knowledge of the secret session ticket key 109 to exploit, in which case you'd be SOL either way.) 110 111 - Change bn_nist.c so that it will properly handle input BIGNUMs 112 outside the expected range. 113 114 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 115 builds. 116 117 [Neel Mehta, Bodo Moeller] 118 119 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 120 the load fails. Useful for distros. 121 [Ben Laurie and the FreeBSD team] 122 123 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 124 [Steve Henson] 125 126 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 127 [Huang Ying] 128 129 *) Expand ENGINE to support engine supplied SSL client certificate functions. 130 131 This work was sponsored by Logica. 132 [Steve Henson] 133 134 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 135 keystores. Support for SSL/TLS client authentication too. 136 Not compiled unless enable-capieng specified to Configure. 137 138 This work was sponsored by Logica. 139 [Steve Henson] 140 141 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 142 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 143 attribute creation routines such as certifcate requests and PKCS#12 144 files. 145 [Steve Henson] 146 147 Changes between 0.9.8g and 0.9.8h [28 May 2008] 148 149 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 150 handshake which could lead to a cilent crash as found using the 151 Codenomicon TLS test suite (CVE-2008-1672) 152 [Steve Henson, Mark Cox] 153 154 *) Fix double free in TLS server name extensions which could lead to 155 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 156 [Joe Orton] 157 158 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 159 160 Clear the error queue to ensure that error entries left from 161 older function calls do not interfere with the correct operation. 162 [Lutz Jaenicke, Erik de Castro Lopo] 163 164 *) Remove root CA certificates of commercial CAs: 165 166 The OpenSSL project does not recommend any specific CA and does not 167 have any policy with respect to including or excluding any CA. 168 Therefore it does not make any sense to ship an arbitrary selection 169 of root CA certificates with the OpenSSL software. 170 [Lutz Jaenicke] 171 172 *) RSA OAEP patches to fix two separate invalid memory reads. 173 The first one involves inputs when 'lzero' is greater than 174 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 175 before the beginning of from). The second one involves inputs where 176 the 'db' section contains nothing but zeroes (there is a one-byte 177 invalid read after the end of 'db'). 178 [Ivan Nestlerode <inestlerode@us.ibm.com>] 179 180 *) Partial backport from 0.9.9-dev: 181 182 Introduce bn_mul_mont (dedicated Montgomery multiplication 183 procedure) as a candidate for BIGNUM assembler implementation. 184 While 0.9.9-dev uses assembler for various architectures, only 185 x86_64 is available by default here in the 0.9.8 branch, and 186 32-bit x86 is available through a compile-time setting. 187 188 To try the 32-bit x86 assembler implementation, use Configure 189 option "enable-montasm" (which exists only for this backport). 190 191 As "enable-montasm" for 32-bit x86 disclaims code stability 192 anyway, in this constellation we activate additional code 193 backported from 0.9.9-dev for further performance improvements, 194 namely BN_from_montgomery_word. (To enable this otherwise, 195 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 196 197 [Andy Polyakov (backport partially by Bodo Moeller)] 198 199 *) Add TLS session ticket callback. This allows an application to set 200 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 201 values. This is useful for key rollover for example where several key 202 sets may exist with different names. 203 [Steve Henson] 204 205 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 206 This was broken until now in 0.9.8 releases, such that the only way 207 a registered ENGINE could be used (assuming it initialises 208 successfully on the host) was to explicitly set it as the default 209 for the relevant algorithms. This is in contradiction with 0.9.7 210 behaviour and the documentation. With this fix, when an ENGINE is 211 registered into a given algorithm's table of implementations, the 212 'uptodate' flag is reset so that auto-discovery will be used next 213 time a new context for that algorithm attempts to select an 214 implementation. 215 [Ian Lister (tweaked by Geoff Thorpe)] 216 217 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 218 implemention in the following ways: 219 220 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 221 hard coded. 222 223 Lack of BER streaming support means one pass streaming processing is 224 only supported if data is detached: setting the streaming flag is 225 ignored for embedded content. 226 227 CMS support is disabled by default and must be explicitly enabled 228 with the enable-cms configuration option. 229 [Steve Henson] 230 231 *) Update the GMP engine glue to do direct copies between BIGNUM and 232 mpz_t when openssl and GMP use the same limb size. Otherwise the 233 existing "conversion via a text string export" trick is still used. 234 [Paul Sheer <paulsheer@gmail.com>] 235 236 *) Zlib compression BIO. This is a filter BIO which compressed and 237 uncompresses any data passed through it. 238 [Steve Henson] 239 240 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 241 RFC3394 compatible AES key wrapping. 242 [Steve Henson] 243 244 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 245 sets string data without copying. X509_ALGOR_set0() and 246 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 247 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 248 from an X509_ATTRIBUTE structure optionally checking it occurs only 249 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 250 data. 251 [Steve Henson] 252 253 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 254 to get the expected BN_FLG_CONSTTIME behavior. 255 [Bodo Moeller (Google)] 256 257 *) Netware support: 258 259 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 260 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 261 - added some more tests to do_tests.pl 262 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 263 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 264 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 265 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 266 - various changes to netware.pl to enable gcc-cross builds on Win32 267 platform 268 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 269 - various changes to fix missing prototype warnings 270 - fixed x86nasm.pl to create correct asm files for NASM COFF output 271 - added AES, WHIRLPOOL and CPUID assembler code to build files 272 - added missing AES assembler make rules to mk1mf.pl 273 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 274 [Guenter Knauf <eflash@gmx.net>] 275 276 *) Implement certificate status request TLS extension defined in RFC3546. 277 A client can set the appropriate parameters and receive the encoded 278 OCSP response via a callback. A server can query the supplied parameters 279 and set the encoded OCSP response in the callback. Add simplified examples 280 to s_client and s_server. 281 [Steve Henson] 282 283 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 284 285 *) Fix various bugs: 286 + Binary incompatibility of ssl_ctx_st structure 287 + DTLS interoperation with non-compliant servers 288 + Don't call get_session_cb() without proposed session 289 + Fix ia64 assembler code 290 [Andy Polyakov, Steve Henson] 291 292 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 293 294 *) DTLS Handshake overhaul. There were longstanding issues with 295 OpenSSL DTLS implementation, which were making it impossible for 296 RFC 4347 compliant client to communicate with OpenSSL server. 297 Unfortunately just fixing these incompatibilities would "cut off" 298 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 299 server keeps tolerating non RFC compliant syntax. The opposite is 300 not true, 0.9.8f client can not communicate with earlier server. 301 This update even addresses CVE-2007-4995. 302 [Andy Polyakov] 303 304 *) Changes to avoid need for function casts in OpenSSL: some compilers 305 (gcc 4.2 and later) reject their use. 306 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 307 Steve Henson] 308 309 *) Add RFC4507 support to OpenSSL. This includes the corrections in 310 RFC4507bis. The encrypted ticket format is an encrypted encoded 311 SSL_SESSION structure, that way new session features are automatically 312 supported. 313 314 If a client application caches session in an SSL_SESSION structure 315 support is transparent because tickets are now stored in the encoded 316 SSL_SESSION. 317 318 The SSL_CTX structure automatically generates keys for ticket 319 protection in servers so again support should be possible 320 with no application modification. 321 322 If a client or server wishes to disable RFC4507 support then the option 323 SSL_OP_NO_TICKET can be set. 324 325 Add a TLS extension debugging callback to allow the contents of any client 326 or server extensions to be examined. 327 328 This work was sponsored by Google. 329 [Steve Henson] 330 331 *) Add initial support for TLS extensions, specifically for the server_name 332 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 333 have new members for a host name. The SSL data structure has an 334 additional member SSL_CTX *initial_ctx so that new sessions can be 335 stored in that context to allow for session resumption, even after the 336 SSL has been switched to a new SSL_CTX in reaction to a client's 337 server_name extension. 338 339 New functions (subject to change): 340 341 SSL_get_servername() 342 SSL_get_servername_type() 343 SSL_set_SSL_CTX() 344 345 New CTRL codes and macros (subject to change): 346 347 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 348 - SSL_CTX_set_tlsext_servername_callback() 349 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 350 - SSL_CTX_set_tlsext_servername_arg() 351 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 352 353 openssl s_client has a new '-servername ...' option. 354 355 openssl s_server has new options '-servername_host ...', '-cert2 ...', 356 '-key2 ...', '-servername_fatal' (subject to change). This allows 357 testing the HostName extension for a specific single host name ('-cert' 358 and '-key' remain fallbacks for handshakes without HostName 359 negotiation). If the unrecogninzed_name alert has to be sent, this by 360 default is a warning; it becomes fatal with the '-servername_fatal' 361 option. 362 363 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 364 365 *) Add AES and SSE2 assembly language support to VC++ build. 366 [Steve Henson] 367 368 *) Mitigate attack on final subtraction in Montgomery reduction. 369 [Andy Polyakov] 370 371 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 372 (which previously caused an internal error). 373 [Bodo Moeller] 374 375 *) Squeeze another 10% out of IGE mode when in != out. 376 [Ben Laurie] 377 378 *) AES IGE mode speedup. 379 [Dean Gaudet (Google)] 380 381 *) Add the Korean symmetric 128-bit cipher SEED (see 382 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 383 add SEED ciphersuites from RFC 4162: 384 385 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 386 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 387 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 388 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 389 390 To minimize changes between patchlevels in the OpenSSL 0.9.8 391 series, SEED remains excluded from compilation unless OpenSSL 392 is configured with 'enable-seed'. 393 [KISA, Bodo Moeller] 394 395 *) Mitigate branch prediction attacks, which can be practical if a 396 single processor is shared, allowing a spy process to extract 397 information. For detailed background information, see 398 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 399 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 400 and Necessary Software Countermeasures"). The core of the change 401 are new versions BN_div_no_branch() and 402 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 403 respectively, which are slower, but avoid the security-relevant 404 conditional branches. These are automatically called by BN_div() 405 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 406 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 407 remove a conditional branch. 408 409 BN_FLG_CONSTTIME is the new name for the previous 410 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 411 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 412 in the exponent causes BN_mod_exp_mont() to use the alternative 413 implementation in BN_mod_exp_mont_consttime().) The old name 414 remains as a deprecated alias. 415 416 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 417 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 418 constant-time implementations for more than just exponentiation. 419 Here too the old name is kept as a deprecated alias. 420 421 BN_BLINDING_new() will now use BN_dup() for the modulus so that 422 the BN_BLINDING structure gets an independent copy of the 423 modulus. This means that the previous "BIGNUM *m" argument to 424 BN_BLINDING_new() and to BN_BLINDING_create_param() now 425 essentially becomes "const BIGNUM *m", although we can't actually 426 change this in the header file before 0.9.9. It allows 427 RSA_setup_blinding() to use BN_with_flags() on the modulus to 428 enable BN_FLG_CONSTTIME. 429 430 [Matthew D Wood (Intel Corp)] 431 432 *) In the SSL/TLS server implementation, be strict about session ID 433 context matching (which matters if an application uses a single 434 external cache for different purposes). Previously, 435 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 436 set. This did ensure strict client verification, but meant that, 437 with applications using a single external cache for quite 438 different requirements, clients could circumvent ciphersuite 439 restrictions for a given session ID context by starting a session 440 in a different context. 441 [Bodo Moeller] 442 443 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 444 a ciphersuite string such as "DEFAULT:RSA" cannot enable 445 authentication-only ciphersuites. 446 [Bodo Moeller] 447 448 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 449 not complete and could lead to a possible single byte overflow 450 (CVE-2007-5135) [Ben Laurie] 451 452 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 453 454 *) Since AES128 and AES256 (and similarly Camellia128 and 455 Camellia256) share a single mask bit in the logic of 456 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 457 kludge to work properly if AES128 is available and AES256 isn't 458 (or if Camellia128 is available and Camellia256 isn't). 459 [Victor Duchovni] 460 461 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 462 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 463 When a point or a seed is encoded in a BIT STRING, we need to 464 prevent the removal of trailing zero bits to get the proper DER 465 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 466 of a NamedBitList, for which trailing 0 bits need to be removed.) 467 [Bodo Moeller] 468 469 *) Have SSL/TLS server implementation tolerate "mismatched" record 470 protocol version while receiving ClientHello even if the 471 ClientHello is fragmented. (The server can't insist on the 472 particular protocol version it has chosen before the ServerHello 473 message has informed the client about his choice.) 474 [Bodo Moeller] 475 476 *) Add RFC 3779 support. 477 [Rob Austein for ARIN, Ben Laurie] 478 479 *) Load error codes if they are not already present instead of using a 480 static variable. This allows them to be cleanly unloaded and reloaded. 481 Improve header file function name parsing. 482 [Steve Henson] 483 484 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 485 or CAPABILITY handshake as required by RFCs. 486 [Goetz Babin-Ebell] 487 488 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 489 490 *) Introduce limits to prevent malicious keys being able to 491 cause a denial of service. (CVE-2006-2940) 492 [Steve Henson, Bodo Moeller] 493 494 *) Fix ASN.1 parsing of certain invalid structures that can result 495 in a denial of service. (CVE-2006-2937) [Steve Henson] 496 497 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 498 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 499 500 *) Fix SSL client code which could crash if connecting to a 501 malicious SSLv2 server. (CVE-2006-4343) 502 [Tavis Ormandy and Will Drewry, Google Security Team] 503 504 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 505 match only those. Before that, "AES256-SHA" would be interpreted 506 as a pattern and match "AES128-SHA" too (since AES128-SHA got 507 the same strength classification in 0.9.7h) as we currently only 508 have a single AES bit in the ciphersuite description bitmap. 509 That change, however, also applied to ciphersuite strings such as 510 "RC4-MD5" that intentionally matched multiple ciphersuites -- 511 namely, SSL 2.0 ciphersuites in addition to the more common ones 512 from SSL 3.0/TLS 1.0. 513 514 So we change the selection algorithm again: Naming an explicit 515 ciphersuite selects this one ciphersuite, and any other similar 516 ciphersuite (same bitmap) from *other* protocol versions. 517 Thus, "RC4-MD5" again will properly select both the SSL 2.0 518 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 519 520 Since SSL 2.0 does not have any ciphersuites for which the 521 128/256 bit distinction would be relevant, this works for now. 522 The proper fix will be to use different bits for AES128 and 523 AES256, which would have avoided the problems from the beginning; 524 however, bits are scarce, so we can only do this in a new release 525 (not just a patchlevel) when we can change the SSL_CIPHER 526 definition to split the single 'unsigned long mask' bitmap into 527 multiple values to extend the available space. 528 529 [Bodo Moeller] 530 531 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 532 533 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 534 (CVE-2006-4339) [Ben Laurie and Google Security Team] 535 536 *) Add AES IGE and biIGE modes. 537 [Ben Laurie] 538 539 *) Change the Unix randomness entropy gathering to use poll() when 540 possible instead of select(), since the latter has some 541 undesirable limitations. 542 [Darryl Miles via Richard Levitte and Bodo Moeller] 543 544 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 545 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 546 cannot be implicitly activated as part of, e.g., the "AES" alias. 547 However, please upgrade to OpenSSL 0.9.9[-dev] for 548 non-experimental use of the ECC ciphersuites to get TLS extension 549 support, which is required for curve and point format negotiation 550 to avoid potential handshake problems. 551 [Bodo Moeller] 552 553 *) Disable rogue ciphersuites: 554 555 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 556 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 557 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 558 559 The latter two were purportedly from 560 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 561 appear there. 562 563 Also deactivate the remaining ciphersuites from 564 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 565 unofficial, and the ID has long expired. 566 [Bodo Moeller] 567 568 *) Fix RSA blinding Heisenbug (problems sometimes occured on 569 dual-core machines) and other potential thread-safety issues. 570 [Bodo Moeller] 571 572 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 573 versions), which is now available for royalty-free use 574 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 575 Also, add Camellia TLS ciphersuites from RFC 4132. 576 577 To minimize changes between patchlevels in the OpenSSL 0.9.8 578 series, Camellia remains excluded from compilation unless OpenSSL 579 is configured with 'enable-camellia'. 580 [NTT] 581 582 *) Disable the padding bug check when compression is in use. The padding 583 bug check assumes the first packet is of even length, this is not 584 necessarily true if compresssion is enabled and can result in false 585 positives causing handshake failure. The actual bug test is ancient 586 code so it is hoped that implementations will either have fixed it by 587 now or any which still have the bug do not support compression. 588 [Steve Henson] 589 590 Changes between 0.9.8a and 0.9.8b [04 May 2006] 591 592 *) When applying a cipher rule check to see if string match is an explicit 593 cipher suite and only match that one cipher suite if it is. 594 [Steve Henson] 595 596 *) Link in manifests for VC++ if needed. 597 [Austin Ziegler <halostatue@gmail.com>] 598 599 *) Update support for ECC-based TLS ciphersuites according to 600 draft-ietf-tls-ecc-12.txt with proposed changes (but without 601 TLS extensions, which are supported starting with the 0.9.9 602 branch, not in the OpenSSL 0.9.8 branch). 603 [Douglas Stebila] 604 605 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 606 opaque EVP_CIPHER_CTX handling. 607 [Steve Henson] 608 609 *) Fixes and enhancements to zlib compression code. We now only use 610 "zlib1.dll" and use the default __cdecl calling convention on Win32 611 to conform with the standards mentioned here: 612 http://www.zlib.net/DLL_FAQ.txt 613 Static zlib linking now works on Windows and the new --with-zlib-include 614 --with-zlib-lib options to Configure can be used to supply the location 615 of the headers and library. Gracefully handle case where zlib library 616 can't be loaded. 617 [Steve Henson] 618 619 *) Several fixes and enhancements to the OID generation code. The old code 620 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 621 handle numbers larger than ULONG_MAX, truncated printing and had a 622 non standard OBJ_obj2txt() behaviour. 623 [Steve Henson] 624 625 *) Add support for building of engines under engine/ as shared libraries 626 under VC++ build system. 627 [Steve Henson] 628 629 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 630 Hopefully, we will not see any false combination of paths any more. 631 [Richard Levitte] 632 633 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 634 635 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 636 (part of SSL_OP_ALL). This option used to disable the 637 countermeasure against man-in-the-middle protocol-version 638 rollback in the SSL 2.0 server implementation, which is a bad 639 idea. (CVE-2005-2969) 640 641 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 642 for Information Security, National Institute of Advanced Industrial 643 Science and Technology [AIST], Japan)] 644 645 *) Add two function to clear and return the verify parameter flags. 646 [Steve Henson] 647 648 *) Keep cipherlists sorted in the source instead of sorting them at 649 runtime, thus removing the need for a lock. 650 [Nils Larsch] 651 652 *) Avoid some small subgroup attacks in Diffie-Hellman. 653 [Nick Mathewson and Ben Laurie] 654 655 *) Add functions for well-known primes. 656 [Nick Mathewson] 657 658 *) Extended Windows CE support. 659 [Satoshi Nakamura and Andy Polyakov] 660 661 *) Initialize SSL_METHOD structures at compile time instead of during 662 runtime, thus removing the need for a lock. 663 [Steve Henson] 664 665 *) Make PKCS7_decrypt() work even if no certificate is supplied by 666 attempting to decrypt each encrypted key in turn. Add support to 667 smime utility. 668 [Steve Henson] 669 670 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 671 672 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 673 OpenSSL 0.9.8.] 674 675 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 676 [Richard Levitte] 677 678 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 679 key into the same file any more. 680 [Richard Levitte] 681 682 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 683 [Andy Polyakov] 684 685 *) Add -utf8 command line and config file option to 'ca'. 686 [Stefan <stf@udoma.org] 687 688 *) Removed the macro des_crypt(), as it seems to conflict with some 689 libraries. Use DES_crypt(). 690 [Richard Levitte] 691 692 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 693 involves renaming the source and generated shared-libs for 694 both. The engines will accept the corrected or legacy ids 695 ('ncipher' and '4758_cca' respectively) when binding. NB, 696 this only applies when building 'shared'. 697 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 698 699 *) Add attribute functions to EVP_PKEY structure. Modify 700 PKCS12_create() to recognize a CSP name attribute and 701 use it. Make -CSP option work again in pkcs12 utility. 702 [Steve Henson] 703 704 *) Add new functionality to the bn blinding code: 705 - automatic re-creation of the BN_BLINDING parameters after 706 a fixed number of uses (currently 32) 707 - add new function for parameter creation 708 - introduce flags to control the update behaviour of the 709 BN_BLINDING parameters 710 - hide BN_BLINDING structure 711 Add a second BN_BLINDING slot to the RSA structure to improve 712 performance when a single RSA object is shared among several 713 threads. 714 [Nils Larsch] 715 716 *) Add support for DTLS. 717 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 718 719 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 720 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 721 [Walter Goulet] 722 723 *) Remove buggy and incompletet DH cert support from 724 ssl/ssl_rsa.c and ssl/s3_both.c 725 [Nils Larsch] 726 727 *) Use SHA-1 instead of MD5 as the default digest algorithm for 728 the apps/openssl applications. 729 [Nils Larsch] 730 731 *) Compile clean with "-Wall -Wmissing-prototypes 732 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 733 DEBUG_SAFESTACK must also be set. 734 [Ben Laurie] 735 736 *) Change ./Configure so that certain algorithms can be disabled by default. 737 The new counterpiece to "no-xxx" is "enable-xxx". 738 739 The patented RC5 and MDC2 algorithms will now be disabled unless 740 "enable-rc5" and "enable-mdc2", respectively, are specified. 741 742 (IDEA remains enabled despite being patented. This is because IDEA 743 is frequently required for interoperability, and there is no license 744 fee for non-commercial use. As before, "no-idea" can be used to 745 avoid this algorithm.) 746 747 [Bodo Moeller] 748 749 *) Add processing of proxy certificates (see RFC 3820). This work was 750 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 751 EGEE (Enabling Grids for E-science in Europe). 752 [Richard Levitte] 753 754 *) RC4 performance overhaul on modern architectures/implementations, such 755 as Intel P4, IA-64 and AMD64. 756 [Andy Polyakov] 757 758 *) New utility extract-section.pl. This can be used specify an alternative 759 section number in a pod file instead of having to treat each file as 760 a separate case in Makefile. This can be done by adding two lines to the 761 pod file: 762 763 =for comment openssl_section:XXX 764 765 The blank line is mandatory. 766 767 [Steve Henson] 768 769 *) New arguments -certform, -keyform and -pass for s_client and s_server 770 to allow alternative format key and certificate files and passphrase 771 sources. 772 [Steve Henson] 773 774 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 775 update associated structures and add various utility functions. 776 777 Add new policy related verify parameters, include policy checking in 778 standard verify code. Enhance 'smime' application with extra parameters 779 to support policy checking and print out. 780 [Steve Henson] 781 782 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 783 Nehemiah processors. These extensions support AES encryption in hardware 784 as well as RNG (though RNG support is currently disabled). 785 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 786 787 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 788 [Geoff Thorpe] 789 790 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 791 [Andy Polyakov and a number of other people] 792 793 *) Improved PowerPC platform support. Most notably BIGNUM assembler 794 implementation contributed by IBM. 795 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 796 797 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 798 exponent rather than 'unsigned long'. There is a corresponding change to 799 the new 'rsa_keygen' element of the RSA_METHOD structure. 800 [Jelte Jansen, Geoff Thorpe] 801 802 *) Functionality for creating the initial serial number file is now 803 moved from CA.pl to the 'ca' utility with a new option -create_serial. 804 805 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 806 number file to 1, which is bound to cause problems. To avoid 807 the problems while respecting compatibility between different 0.9.7 808 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 809 CA.pl for serial number initialization. With the new release 0.9.8, 810 we can fix the problem directly in the 'ca' utility.) 811 [Steve Henson] 812 813 *) Reduced header interdepencies by declaring more opaque objects in 814 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 815 give fewer recursive includes, which could break lazy source code - so 816 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 817 developers should define this symbol when building and using openssl to 818 ensure they track the recommended behaviour, interfaces, [etc], but 819 backwards-compatible behaviour prevails when this isn't defined. 820 [Geoff Thorpe] 821 822 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 823 [Steve Henson] 824 825 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 826 This will generate a random key of the appropriate length based on the 827 cipher context. The EVP_CIPHER can provide its own random key generation 828 routine to support keys of a specific form. This is used in the des and 829 3des routines to generate a key of the correct parity. Update S/MIME 830 code to use new functions and hence generate correct parity DES keys. 831 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 832 valid (weak or incorrect parity). 833 [Steve Henson] 834 835 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 836 as looking them up. This is useful when the verified structure may contain 837 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 838 present unless the new PKCS7_NO_CRL flag is asserted. 839 [Steve Henson] 840 841 *) Extend ASN1 oid configuration module. It now additionally accepts the 842 syntax: 843 844 shortName = some long name, 1.2.3.4 845 [Steve Henson] 846 847 *) Reimplemented the BN_CTX implementation. There is now no more static 848 limitation on the number of variables it can handle nor the depth of the 849 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 850 information can now expand as required, and rather than having a single 851 static array of bignums, BN_CTX now uses a linked-list of such arrays 852 allowing it to expand on demand whilst maintaining the usefulness of 853 BN_CTX's "bundling". 854 [Geoff Thorpe] 855 856 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 857 to allow all RSA operations to function using a single BN_CTX. 858 [Geoff Thorpe] 859 860 *) Preliminary support for certificate policy evaluation and checking. This 861 is initially intended to pass the tests outlined in "Conformance Testing 862 of Relying Party Client Certificate Path Processing Logic" v1.07. 863 [Steve Henson] 864 865 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 866 remained unused and not that useful. A variety of other little bignum 867 tweaks and fixes have also been made continuing on from the audit (see 868 below). 869 [Geoff Thorpe] 870 871 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 872 associated ASN1, EVP and SSL functions and old ASN1 macros. 873 [Richard Levitte] 874 875 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 876 and this should never fail. So the return value from the use of 877 BN_set_word() (which can fail due to needless expansion) is now deprecated; 878 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 879 [Geoff Thorpe] 880 881 *) BN_CTX_get() should return zero-valued bignums, providing the same 882 initialised value as BN_new(). 883 [Geoff Thorpe, suggested by Ulf M�ller] 884 885 *) Support for inhibitAnyPolicy certificate extension. 886 [Steve Henson] 887 888 *) An audit of the BIGNUM code is underway, for which debugging code is 889 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 890 is considered valid when processing BIGNUMs, and causes execution to 891 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 892 further steps are taken to deliberately pollute unused data in BIGNUM 893 structures to try and expose faulty code further on. For now, openssl will 894 (in its default mode of operation) continue to tolerate the inconsistent 895 forms that it has tolerated in the past, but authors and packagers should 896 consider trying openssl and their own applications when compiled with 897 these debugging symbols defined. It will help highlight potential bugs in 898 their own code, and will improve the test coverage for OpenSSL itself. At 899 some point, these tighter rules will become openssl's default to improve 900 maintainability, though the assert()s and other overheads will remain only 901 in debugging configurations. See bn.h for more details. 902 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 903 904 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 905 that can only be obtained through BN_CTX_new() (which implicitly 906 initialises it). The presence of this function only made it possible 907 to overwrite an existing structure (and cause memory leaks). 908 [Geoff Thorpe] 909 910 *) Because of the callback-based approach for implementing LHASH as a 911 template type, lh_insert() adds opaque objects to hash-tables and 912 lh_doall() or lh_doall_arg() are typically used with a destructor callback 913 to clean up those corresponding objects before destroying the hash table 914 (and losing the object pointers). So some over-zealous constifications in 915 LHASH have been relaxed so that lh_insert() does not take (nor store) the 916 objects as "const" and the lh_doall[_arg] callback wrappers are not 917 prototyped to have "const" restrictions on the object pointers they are 918 given (and so aren't required to cast them away any more). 919 [Geoff Thorpe] 920 921 *) The tmdiff.h API was so ugly and minimal that our own timing utility 922 (speed) prefers to use its own implementation. The two implementations 923 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 924 its object type properly exposed (MS_TM) instead of casting to/from "char 925 *". This may still change yet if someone realises MS_TM and "ms_time_***" 926 aren't necessarily the greatest nomenclatures - but this is what was used 927 internally to the implementation so I've used that for now. 928 [Geoff Thorpe] 929 930 *) Ensure that deprecated functions do not get compiled when 931 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 932 the self-tests were still using deprecated key-generation functions so 933 these have been updated also. 934 [Geoff Thorpe] 935 936 *) Reorganise PKCS#7 code to separate the digest location functionality 937 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 938 New function PKCS7_set_digest() to set the digest type for PKCS#7 939 digestedData type. Add additional code to correctly generate the 940 digestedData type and add support for this type in PKCS7 initialization 941 functions. 942 [Steve Henson] 943 944 *) New function PKCS7_set0_type_other() this initializes a PKCS7 945 structure of type "other". 946 [Steve Henson] 947 948 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 949 sure the loop does correctly stop and breaking ("division by zero") 950 modulus operations are not performed. The (pre-generated) prime 951 table crypto/bn/bn_prime.h was already correct, but it could not be 952 re-generated on some platforms because of the "division by zero" 953 situation in the script. 954 [Ralf S. Engelschall] 955 956 *) Update support for ECC-based TLS ciphersuites according to 957 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 958 SHA-1 now is only used for "small" curves (where the 959 representation of a field element takes up to 24 bytes); for 960 larger curves, the field element resulting from ECDH is directly 961 used as premaster secret. 962 [Douglas Stebila (Sun Microsystems Laboratories)] 963 964 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 965 curve secp160r1 to the tests. 966 [Douglas Stebila (Sun Microsystems Laboratories)] 967 968 *) Add the possibility to load symbols globally with DSO. 969 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 970 971 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 972 control of the error stack. 973 [Richard Levitte] 974 975 *) Add support for STORE in ENGINE. 976 [Richard Levitte] 977 978 *) Add the STORE type. The intention is to provide a common interface 979 to certificate and key stores, be they simple file-based stores, or 980 HSM-type store, or LDAP stores, or... 981 NOTE: The code is currently UNTESTED and isn't really used anywhere. 982 [Richard Levitte] 983 984 *) Add a generic structure called OPENSSL_ITEM. This can be used to 985 pass a list of arguments to any function as well as provide a way 986 for a function to pass data back to the caller. 987 [Richard Levitte] 988 989 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 990 works like BUF_strdup() but can be used to duplicate a portion of 991 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 992 a memory area. 993 [Richard Levitte] 994 995 *) Add the function sk_find_ex() which works like sk_find(), but will 996 return an index to an element even if an exact match couldn't be 997 found. The index is guaranteed to point at the element where the 998 searched-for key would be inserted to preserve sorting order. 999 [Richard Levitte] 1000 1001 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 1002 takes an extra flags argument for optional functionality. Currently, 1003 the following flags are defined: 1004 1005 OBJ_BSEARCH_VALUE_ON_NOMATCH 1006 This one gets OBJ_bsearch_ex() to return a pointer to the first 1007 element where the comparing function returns a negative or zero 1008 number. 1009 1010 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 1011 This one gets OBJ_bsearch_ex() to return a pointer to the first 1012 element where the comparing function returns zero. This is useful 1013 if there are more than one element where the comparing function 1014 returns zero. 1015 [Richard Levitte] 1016 1017 *) Make it possible to create self-signed certificates with 'openssl ca' 1018 in such a way that the self-signed certificate becomes part of the 1019 CA database and uses the same mechanisms for serial number generation 1020 as all other certificate signing. The new flag '-selfsign' enables 1021 this functionality. Adapt CA.sh and CA.pl.in. 1022 [Richard Levitte] 1023 1024 *) Add functionality to check the public key of a certificate request 1025 against a given private. This is useful to check that a certificate 1026 request can be signed by that key (self-signing). 1027 [Richard Levitte] 1028 1029 *) Make it possible to have multiple active certificates with the same 1030 subject in the CA index file. This is done only if the keyword 1031 'unique_subject' is set to 'no' in the main CA section (default 1032 if 'CA_default') of the configuration file. The value is saved 1033 with the database itself in a separate index attribute file, 1034 named like the index file with '.attr' appended to the name. 1035 [Richard Levitte] 1036 1037 *) Generate muti valued AVAs using '+' notation in config files for 1038 req and dirName. 1039 [Steve Henson] 1040 1041 *) Support for nameConstraints certificate extension. 1042 [Steve Henson] 1043 1044 *) Support for policyConstraints certificate extension. 1045 [Steve Henson] 1046 1047 *) Support for policyMappings certificate extension. 1048 [Steve Henson] 1049 1050 *) Make sure the default DSA_METHOD implementation only uses its 1051 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 1052 and change its own handlers to be NULL so as to remove unnecessary 1053 indirection. This lets alternative implementations fallback to the 1054 default implementation more easily. 1055 [Geoff Thorpe] 1056 1057 *) Support for directoryName in GeneralName related extensions 1058 in config files. 1059 [Steve Henson] 1060 1061 *) Make it possible to link applications using Makefile.shared. 1062 Make that possible even when linking against static libraries! 1063 [Richard Levitte] 1064 1065 *) Support for single pass processing for S/MIME signing. This now 1066 means that S/MIME signing can be done from a pipe, in addition 1067 cleartext signing (multipart/signed type) is effectively streaming 1068 and the signed data does not need to be all held in memory. 1069 1070 This is done with a new flag PKCS7_STREAM. When this flag is set 1071 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 1072 is done after the data is output (and digests calculated) in 1073 SMIME_write_PKCS7(). 1074 [Steve Henson] 1075 1076 *) Add full support for -rpath/-R, both in shared libraries and 1077 applications, at least on the platforms where it's known how 1078 to do it. 1079 [Richard Levitte] 1080 1081 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 1082 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 1083 will now compute a table of multiples of the generator that 1084 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 1085 faster (notably in the case of a single point multiplication, 1086 scalar * generator). 1087 [Nils Larsch, Bodo Moeller] 1088 1089 *) IPv6 support for certificate extensions. The various extensions 1090 which use the IP:a.b.c.d can now take IPv6 addresses using the 1091 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 1092 correctly. 1093 [Steve Henson] 1094 1095 *) Added an ENGINE that implements RSA by performing private key 1096 exponentiations with the GMP library. The conversions to and from 1097 GMP's mpz_t format aren't optimised nor are any montgomery forms 1098 cached, and on x86 it appears OpenSSL's own performance has caught up. 1099 However there are likely to be other architectures where GMP could 1100 provide a boost. This ENGINE is not built in by default, but it can be 1101 specified at Configure time and should be accompanied by the necessary 1102 linker additions, eg; 1103 ./config -DOPENSSL_USE_GMP -lgmp 1104 [Geoff Thorpe] 1105 1106 *) "openssl engine" will not display ENGINE/DSO load failure errors when 1107 testing availability of engines with "-t" - the old behaviour is 1108 produced by increasing the feature's verbosity with "-tt". 1109 [Geoff Thorpe] 1110 1111 *) ECDSA routines: under certain error conditions uninitialized BN objects 1112 could be freed. Solution: make sure initialization is performed early 1113 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 1114 via PR#459) 1115 [Lutz Jaenicke] 1116 1117 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 1118 and DH_METHOD (eg. by ENGINE implementations) to override the normal 1119 software implementations. For DSA and DH, parameter generation can 1120 also be overriden by providing the appropriate method callbacks. 1121 [Geoff Thorpe] 1122 1123 *) Change the "progress" mechanism used in key-generation and 1124 primality testing to functions that take a new BN_GENCB pointer in 1125 place of callback/argument pairs. The new API functions have "_ex" 1126 postfixes and the older functions are reimplemented as wrappers for 1127 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 1128 declarations of the old functions to help (graceful) attempts to 1129 migrate to the new functions. Also, the new key-generation API 1130 functions operate on a caller-supplied key-structure and return 1131 success/failure rather than returning a key or NULL - this is to 1132 help make "keygen" another member function of RSA_METHOD etc. 1133 1134 Example for using the new callback interface: 1135 1136 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 1137 void *my_arg = ...; 1138 BN_GENCB my_cb; 1139 1140 BN_GENCB_set(&my_cb, my_callback, my_arg); 1141 1142 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 1143 /* For the meaning of a, b in calls to my_callback(), see the 1144 * documentation of the function that calls the callback. 1145 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 1146 * my_callback should return 1 if it wants BN_is_prime_ex() 1147 * to continue, or 0 to stop. 1148 */ 1149 1150 [Geoff Thorpe] 1151 1152 *) Change the ZLIB compression method to be stateful, and make it 1153 available to TLS with the number defined in 1154 draft-ietf-tls-compression-04.txt. 1155 [Richard Levitte] 1156 1157 *) Add the ASN.1 structures and functions for CertificatePair, which 1158 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 1159 1160 CertificatePair ::= SEQUENCE { 1161 forward [0] Certificate OPTIONAL, 1162 reverse [1] Certificate OPTIONAL, 1163 -- at least one of the pair shall be present -- } 1164 1165 Also implement the PEM functions to read and write certificate 1166 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 1167 1168 This needed to be defined, mostly for the sake of the LDAP 1169 attribute crossCertificatePair, but may prove useful elsewhere as 1170 well. 1171 [Richard Levitte] 1172 1173 *) Make it possible to inhibit symlinking of shared libraries in 1174 Makefile.shared, for Cygwin's sake. 1175 [Richard Levitte] 1176 1177 *) Extend the BIGNUM API by creating a function 1178 void BN_set_negative(BIGNUM *a, int neg); 1179 and a macro that behave like 1180 int BN_is_negative(const BIGNUM *a); 1181 1182 to avoid the need to access 'a->neg' directly in applications. 1183 [Nils Larsch] 1184 1185 *) Implement fast modular reduction for pseudo-Mersenne primes 1186 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 1187 EC_GROUP_new_curve_GFp() will now automatically use this 1188 if applicable. 1189 [Nils Larsch <nla@trustcenter.de>] 1190 1191 *) Add new lock type (CRYPTO_LOCK_BN). 1192 [Bodo Moeller] 1193 1194 *) Change the ENGINE framework to automatically load engines 1195 dynamically from specific directories unless they could be 1196 found to already be built in or loaded. Move all the 1197 current engines except for the cryptodev one to a new 1198 directory engines/. 1199 The engines in engines/ are built as shared libraries if 1200 the "shared" options was given to ./Configure or ./config. 1201 Otherwise, they are inserted in libcrypto.a. 1202 /usr/local/ssl/engines is the default directory for dynamic 1203 engines, but that can be overriden at configure time through 1204 the usual use of --prefix and/or --openssldir, and at run 1205 time with the environment variable OPENSSL_ENGINES. 1206 [Geoff Thorpe and Richard Levitte] 1207 1208 *) Add Makefile.shared, a helper makefile to build shared 1209 libraries. Addapt Makefile.org. 1210 [Richard Levitte] 1211 1212 *) Add version info to Win32 DLLs. 1213 [Peter 'Luna' Runestig" <peter@runestig.com>] 1214 1215 *) Add new 'medium level' PKCS#12 API. Certificates and keys 1216 can be added using this API to created arbitrary PKCS#12 1217 files while avoiding the low level API. 1218 1219 New options to PKCS12_create(), key or cert can be NULL and 1220 will then be omitted from the output file. The encryption 1221 algorithm NIDs can be set to -1 for no encryption, the mac 1222 iteration count can be set to 0 to omit the mac. 1223 1224 Enhance pkcs12 utility by making the -nokeys and -nocerts 1225 options work when creating a PKCS#12 file. New option -nomac 1226 to omit the mac, NONE can be set for an encryption algorithm. 1227 New code is modified to use the enhanced PKCS12_create() 1228 instead of the low level API. 1229 [Steve Henson] 1230 1231 *) Extend ASN1 encoder to support indefinite length constructed 1232 encoding. This can output sequences tags and octet strings in 1233 this form. Modify pk7_asn1.c to support indefinite length 1234 encoding. This is experimental and needs additional code to 1235 be useful, such as an ASN1 bio and some enhanced streaming 1236 PKCS#7 code. 1237 1238 Extend template encode functionality so that tagging is passed 1239 down to the template encoder. 1240 [Steve Henson] 1241 1242 *) Let 'openssl req' fail if an argument to '-newkey' is not 1243 recognized instead of using RSA as a default. 1244 [Bodo Moeller] 1245 1246 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 1247 As these are not official, they are not included in "ALL"; 1248 the "ECCdraft" ciphersuite group alias can be used to select them. 1249 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 1250 1251 *) Add ECDH engine support. 1252 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 1253 1254 *) Add ECDH in new directory crypto/ecdh/. 1255 [Douglas Stebila (Sun Microsystems Laboratories)] 1256 1257 *) Let BN_rand_range() abort with an error after 100 iterations 1258 without success (which indicates a broken PRNG). 1259 [Bodo Moeller] 1260 1261 *) Change BN_mod_sqrt() so that it verifies that the input value 1262 is really the square of the return value. (Previously, 1263 BN_mod_sqrt would show GIGO behaviour.) 1264 [Bodo Moeller] 1265 1266 *) Add named elliptic curves over binary fields from X9.62, SECG, 1267 and WAP/WTLS; add OIDs that were still missing. 1268 1269 [Sheueling Chang Shantz and Douglas Stebila 1270 (Sun Microsystems Laboratories)] 1271 1272 *) Extend the EC library for elliptic curves over binary fields 1273 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 1274 New EC_METHOD: 1275 1276 EC_GF2m_simple_method 1277 1278 New API functions: 1279 1280 EC_GROUP_new_curve_GF2m 1281 EC_GROUP_set_curve_GF2m 1282 EC_GROUP_get_curve_GF2m 1283 EC_POINT_set_affine_coordinates_GF2m 1284 EC_POINT_get_affine_coordinates_GF2m 1285 EC_POINT_set_compressed_coordinates_GF2m 1286 1287 Point compression for binary fields is disabled by default for 1288 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 1289 enable it). 1290 1291 As binary polynomials are represented as BIGNUMs, various members 1292 of the EC_GROUP and EC_POINT data structures can be shared 1293 between the implementations for prime fields and binary fields; 1294 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 1295 are essentially identical to their ..._GFp counterparts. 1296 (For simplicity, the '..._GFp' prefix has been dropped from 1297 various internal method names.) 1298 1299 An internal 'field_div' method (similar to 'field_mul' and 1300 'field_sqr') has been added; this is used only for binary fields. 1301 1302 [Sheueling Chang Shantz and Douglas Stebila 1303 (Sun Microsystems Laboratories)] 1304 1305 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 1306 through methods ('mul', 'precompute_mult'). 1307 1308 The generic implementations (now internally called 'ec_wNAF_mul' 1309 and 'ec_wNAF_precomputed_mult') remain the default if these 1310 methods are undefined. 1311 1312 [Sheueling Chang Shantz and Douglas Stebila 1313 (Sun Microsystems Laboratories)] 1314 1315 *) New function EC_GROUP_get_degree, which is defined through 1316 EC_METHOD. For curves over prime fields, this returns the bit 1317 length of the modulus. 1318 1319 [Sheueling Chang Shantz and Douglas Stebila 1320 (Sun Microsystems Laboratories)] 1321 1322 *) New functions EC_GROUP_dup, EC_POINT_dup. 1323 (These simply call ..._new and ..._copy). 1324 1325 [Sheueling Chang Shantz and Douglas Stebila 1326 (Sun Microsystems Laboratories)] 1327 1328 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 1329 Polynomials are represented as BIGNUMs (where the sign bit is not 1330 used) in the following functions [macros]: 1331 1332 BN_GF2m_add 1333 BN_GF2m_sub [= BN_GF2m_add] 1334 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 1335 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 1336 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 1337 BN_GF2m_mod_inv 1338 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 1339 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 1340 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 1341 BN_GF2m_cmp [= BN_ucmp] 1342 1343 (Note that only the 'mod' functions are actually for fields GF(2^m). 1344 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 1345 1346 For some functions, an the irreducible polynomial defining a 1347 field can be given as an 'unsigned int[]' with strictly 1348 decreasing elements giving the indices of those bits that are set; 1349 i.e., p[] represents the polynomial 1350 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 1351 where 1352 p[0] > p[1] > ... > p[k] = 0. 1353 This applies to the following functions: 1354 1355 BN_GF2m_mod_arr 1356 BN_GF2m_mod_mul_arr 1357 BN_GF2m_mod_sqr_arr 1358 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 1359 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 1360 BN_GF2m_mod_exp_arr 1361 BN_GF2m_mod_sqrt_arr 1362 BN_GF2m_mod_solve_quad_arr 1363 BN_GF2m_poly2arr 1364 BN_GF2m_arr2poly 1365 1366 Conversion can be performed by the following functions: 1367 1368 BN_GF2m_poly2arr 1369 BN_GF2m_arr2poly 1370 1371 bntest.c has additional tests for binary polynomial arithmetic. 1372 1373 Two implementations for BN_GF2m_mod_div() are available. 1374 The default algorithm simply uses BN_GF2m_mod_inv() and 1375 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 1376 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 1377 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 1378 1379 [Sheueling Chang Shantz and Douglas Stebila 1380 (Sun Microsystems Laboratories)] 1381 1382 *) Add new error code 'ERR_R_DISABLED' that can be used when some 1383 functionality is disabled at compile-time. 1384 [Douglas Stebila <douglas.stebila@sun.com>] 1385 1386 *) Change default behaviour of 'openssl asn1parse' so that more 1387 information is visible when viewing, e.g., a certificate: 1388 1389 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 1390 mode the content of non-printable OCTET STRINGs is output in a 1391 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 1392 avoid the appearance of a printable string. 1393 [Nils Larsch <nla@trustcenter.de>] 1394 1395 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 1396 functions 1397 EC_GROUP_set_asn1_flag() 1398 EC_GROUP_get_asn1_flag() 1399 EC_GROUP_set_point_conversion_form() 1400 EC_GROUP_get_point_conversion_form() 1401 These control ASN1 encoding details: 1402 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 1403 has been set to OPENSSL_EC_NAMED_CURVE. 1404 - Points are encoded in uncompressed form by default; options for 1405 asn1_for are as for point2oct, namely 1406 POINT_CONVERSION_COMPRESSED 1407 POINT_CONVERSION_UNCOMPRESSED 1408 POINT_CONVERSION_HYBRID 1409 1410 Also add 'seed' and 'seed_len' members to EC_GROUP with access 1411 functions 1412 EC_GROUP_set_seed() 1413 EC_GROUP_get0_seed() 1414 EC_GROUP_get_seed_len() 1415 This is used only for ASN1 purposes (so far). 1416 [Nils Larsch <nla@trustcenter.de>] 1417 1418 *) Add 'field_type' member to EC_METHOD, which holds the NID 1419 of the appropriate field type OID. The new function 1420 EC_METHOD_get_field_type() returns this value. 1421 [Nils Larsch <nla@trustcenter.de>] 1422 1423 *) Add functions 1424 EC_POINT_point2bn() 1425 EC_POINT_bn2point() 1426 EC_POINT_point2hex() 1427 EC_POINT_hex2point() 1428 providing useful interfaces to EC_POINT_point2oct() and 1429 EC_POINT_oct2point(). 1430 [Nils Larsch <nla@trustcenter.de>] 1431 1432 *) Change internals of the EC library so that the functions 1433 EC_GROUP_set_generator() 1434 EC_GROUP_get_generator() 1435 EC_GROUP_get_order() 1436 EC_GROUP_get_cofactor() 1437 are implemented directly in crypto/ec/ec_lib.c and not dispatched 1438 to methods, which would lead to unnecessary code duplication when 1439 adding different types of curves. 1440 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 1441 1442 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 1443 arithmetic, and such that modified wNAFs are generated 1444 (which avoid length expansion in many cases). 1445 [Bodo Moeller] 1446 1447 *) Add a function EC_GROUP_check_discriminant() (defined via 1448 EC_METHOD) that verifies that the curve discriminant is non-zero. 1449 1450 Add a function EC_GROUP_check() that makes some sanity tests 1451 on a EC_GROUP, its generator and order. This includes 1452 EC_GROUP_check_discriminant(). 1453 [Nils Larsch <nla@trustcenter.de>] 1454 1455 *) Add ECDSA in new directory crypto/ecdsa/. 1456 1457 Add applications 'openssl ecparam' and 'openssl ecdsa' 1458 (these are based on 'openssl dsaparam' and 'openssl dsa'). 1459 1460 ECDSA support is also included in various other files across the 1461 library. Most notably, 1462 - 'openssl req' now has a '-newkey ecdsa:file' option; 1463 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 1464 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 1465 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 1466 them suitable for ECDSA where domain parameters must be 1467 extracted before the specific public key; 1468 - ECDSA engine support has been added. 1469 [Nils Larsch <nla@trustcenter.de>] 1470 1471 *) Include some named elliptic curves, and add OIDs from X9.62, 1472 SECG, and WAP/WTLS. Each curve can be obtained from the new 1473 function 1474 EC_GROUP_new_by_curve_name(), 1475 and the list of available named curves can be obtained with 1476 EC_get_builtin_curves(). 1477 Also add a 'curve_name' member to EC_GROUP objects, which can be 1478 accessed via 1479 EC_GROUP_set_curve_name() 1480 EC_GROUP_get_curve_name() 1481 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 1482 1483 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1484 was actually never needed) and in BN_mul(). The removal in BN_mul() 1485 required a small change in bn_mul_part_recursive() and the addition 1486 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1487 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1488 bn_sub_words() and bn_add_words() except they take arrays with 1489 differing sizes. 1490 [Richard Levitte] 1491 1492 Changes between 0.9.7m and 0.9.7n [xx XXX xxxx] 1493 1494 *) In the SSL/TLS server implementation, be strict about session ID 1495 context matching (which matters if an application uses a single 1496 external cache for different purposes). Previously, 1497 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 1498 set. This did ensure strict client verification, but meant that, 1499 with applications using a single external cache for quite 1500 different requirements, clients could circumvent ciphersuite 1501 restrictions for a given session ID context by starting a session 1502 in a different context. 1503 [Bodo Moeller] 1504 1505 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 1506 1507 *) Cleanse PEM buffers before freeing them since they may contain 1508 sensitive data. 1509 [Benjamin Bennett <ben@psc.edu>] 1510 1511 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 1512 a ciphersuite string such as "DEFAULT:RSA" cannot enable 1513 authentication-only ciphersuites. 1514 [Bodo Moeller] 1515 1516 *) Since AES128 and AES256 share a single mask bit in the logic of 1517 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 1518 kludge to work properly if AES128 is available and AES256 isn't. 1519 [Victor Duchovni] 1520 1521 *) Expand security boundary to match 1.1.1 module. 1522 [Steve Henson] 1523 1524 *) Remove redundant features: hash file source, editing of test vectors 1525 modify fipsld to use external fips_premain.c signature. 1526 [Steve Henson] 1527 1528 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 1529 run algorithm test programs. 1530 [Steve Henson] 1531 1532 *) Make algorithm test programs more tolerant of whitespace. 1533 [Steve Henson] 1534 1535 *) Have SSL/TLS server implementation tolerate "mismatched" record 1536 protocol version while receiving ClientHello even if the 1537 ClientHello is fragmented. (The server can't insist on the 1538 particular protocol version it has chosen before the ServerHello 1539 message has informed the client about his choice.) 1540 [Bodo Moeller] 1541 1542 *) Load error codes if they are not already present instead of using a 1543 static variable. This allows them to be cleanly unloaded and reloaded. 1544 [Steve Henson] 1545 1546 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 1547 1548 *) Introduce limits to prevent malicious keys being able to 1549 cause a denial of service. (CVE-2006-2940) 1550 [Steve Henson, Bodo Moeller] 1551 1552 *) Fix ASN.1 parsing of certain invalid structures that can result 1553 in a denial of service. (CVE-2006-2937) [Steve Henson] 1554 1555 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 1556 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 1557 1558 *) Fix SSL client code which could crash if connecting to a 1559 malicious SSLv2 server. (CVE-2006-4343) 1560 [Tavis Ormandy and Will Drewry, Google Security Team] 1561 1562 *) Change ciphersuite string processing so that an explicit 1563 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 1564 will no longer include "AES128-SHA"), and any other similar 1565 ciphersuite (same bitmap) from *other* protocol versions (so that 1566 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 1567 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 1568 changes from 0.9.8b and 0.9.8d. 1569 [Bodo Moeller] 1570 1571 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 1572 1573 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1574 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1575 1576 *) Change the Unix randomness entropy gathering to use poll() when 1577 possible instead of select(), since the latter has some 1578 undesirable limitations. 1579 [Darryl Miles via Richard Levitte and Bodo Moeller] 1580 1581 *) Disable rogue ciphersuites: 1582 1583 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1584 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1585 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1586 1587 The latter two were purportedly from 1588 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1589 appear there. 1590 1591 Also deactive the remaining ciphersuites from 1592 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1593 unofficial, and the ID has long expired. 1594 [Bodo Moeller] 1595 1596 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1597 dual-core machines) and other potential thread-safety issues. 1598 [Bodo Moeller] 1599 1600 Changes between 0.9.7i and 0.9.7j [04 May 2006] 1601 1602 *) Adapt fipsld and the build system to link against the validated FIPS 1603 module in FIPS mode. 1604 [Steve Henson] 1605 1606 *) Fixes for VC++ 2005 build under Windows. 1607 [Steve Henson] 1608 1609 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 1610 from a Windows bash shell such as MSYS. It is autodetected from the 1611 "config" script when run from a VC++ environment. Modify standard VC++ 1612 build to use fipscanister.o from the GNU make build. 1613 [Steve Henson] 1614 1615 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 1616 1617 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 1618 The value now differs depending on if you build for FIPS or not. 1619 BEWARE! A program linked with a shared FIPSed libcrypto can't be 1620 safely run with a non-FIPSed libcrypto, as it may crash because of 1621 the difference induced by this change. 1622 [Andy Polyakov] 1623 1624 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 1625 1626 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1627 (part of SSL_OP_ALL). This option used to disable the 1628 countermeasure against man-in-the-middle protocol-version 1629 rollback in the SSL 2.0 server implementation, which is a bad 1630 idea. (CVE-2005-2969) 1631 1632 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1633 for Information Security, National Institute of Advanced Industrial 1634 Science and Technology [AIST], Japan)] 1635 1636 *) Minimal support for X9.31 signatures and PSS padding modes. This is 1637 mainly for FIPS compliance and not fully integrated at this stage. 1638 [Steve Henson] 1639 1640 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 1641 the exponentiation using a fixed-length exponent. (Otherwise, 1642 the information leaked through timing could expose the secret key 1643 after many signatures; cf. Bleichenbacher's attack on DSA with 1644 biased k.) 1645 [Bodo Moeller] 1646 1647 *) Make a new fixed-window mod_exp implementation the default for 1648 RSA, DSA, and DH private-key operations so that the sequence of 1649 squares and multiplies and the memory access pattern are 1650 independent of the particular secret key. This will mitigate 1651 cache-timing and potential related attacks. 1652 1653 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 1654 and this is automatically used by BN_mod_exp_mont() if the new flag 1655 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 1656 will use this BN flag for private exponents unless the flag 1657 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 1658 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 1659 1660 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 1661 1662 *) Change the client implementation for SSLv23_method() and 1663 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 1664 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 1665 (Previously, the SSL 2.0 backwards compatible Client Hello 1666 message format would be used even with SSL_OP_NO_SSLv2.) 1667 [Bodo Moeller] 1668 1669 *) Add support for smime-type MIME parameter in S/MIME messages which some 1670 clients need. 1671 [Steve Henson] 1672 1673 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 1674 a threadsafe manner. Modify rsa code to use new function and add calls 1675 to dsa and dh code (which had race conditions before). 1676 [Steve Henson] 1677 1678 *) Include the fixed error library code in the C error file definitions 1679 instead of fixing them up at runtime. This keeps the error code 1680 structures constant. 1681 [Steve Henson] 1682 1683 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 1684 1685 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 1686 OpenSSL 0.9.8.] 1687 1688 *) Fixes for newer kerberos headers. NB: the casts are needed because 1689 the 'length' field is signed on one version and unsigned on another 1690 with no (?) obvious way to tell the difference, without these VC++ 1691 complains. Also the "definition" of FAR (blank) is no longer included 1692 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 1693 some needed definitions. 1694 [Steve Henson] 1695 1696 *) Undo Cygwin change. 1697 [Ulf M�ller] 1698 1699 *) Added support for proxy certificates according to RFC 3820. 1700 Because they may be a security thread to unaware applications, 1701 they must be explicitely allowed in run-time. See 1702 docs/HOWTO/proxy_certificates.txt for further information. 1703 [Richard Levitte] 1704 1705 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 1706 1707 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 1708 server and client random values. Previously 1709 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 1710 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 1711 1712 This change has negligible security impact because: 1713 1714 1. Server and client random values still have 24 bytes of pseudo random 1715 data. 1716 1717 2. Server and client random values are sent in the clear in the initial 1718 handshake. 1719 1720 3. The master secret is derived using the premaster secret (48 bytes in 1721 size for static RSA ciphersuites) as well as client server and random 1722 values. 1723 1724 The OpenSSL team would like to thank the UK NISCC for bringing this issue 1725 to our attention. 1726 1727 [Stephen Henson, reported by UK NISCC] 1728 1729 *) Use Windows randomness collection on Cygwin. 1730 [Ulf M�ller] 1731 1732 *) Fix hang in EGD/PRNGD query when communication socket is closed 1733 prematurely by EGD/PRNGD. 1734 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 1735 1736 *) Prompt for pass phrases when appropriate for PKCS12 input format. 1737 [Steve Henson] 1738 1739 *) Back-port of selected performance improvements from development 1740 branch, as well as improved support for PowerPC platforms. 1741 [Andy Polyakov] 1742 1743 *) Add lots of checks for memory allocation failure, error codes to indicate 1744 failure and freeing up memory if a failure occurs. 1745 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 1746 1747 *) Add new -passin argument to dgst. 1748 [Steve Henson] 1749 1750 *) Perform some character comparisons of different types in X509_NAME_cmp: 1751 this is needed for some certificates that reencode DNs into UTF8Strings 1752 (in violation of RFC3280) and can't or wont issue name rollover 1753 certificates. 1754 [Steve Henson] 1755 1756 *) Make an explicit check during certificate validation to see that 1757 the CA setting in each certificate on the chain is correct. As a 1758 side effect always do the following basic checks on extensions, 1759 not just when there's an associated purpose to the check: 1760 1761 - if there is an unhandled critical extension (unless the user 1762 has chosen to ignore this fault) 1763 - if the path length has been exceeded (if one is set at all) 1764 - that certain extensions fit the associated purpose (if one has 1765 been given) 1766 [Richard Levitte] 1767 1768 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 1769 1770 *) Avoid a race condition when CRLs are checked in a multi threaded 1771 environment. This would happen due to the reordering of the revoked 1772 entries during signature checking and serial number lookup. Now the 1773 encoding is cached and the serial number sort performed under a lock. 1774 Add new STACK function sk_is_sorted(). 1775 [Steve Henson] 1776 1777 *) Add Delta CRL to the extension code. 1778 [Steve Henson] 1779 1780 *) Various fixes to s3_pkt.c so alerts are sent properly. 1781 [David Holmes <d.holmes@f5.com>] 1782 1783 *) Reduce the chances of duplicate issuer name and serial numbers (in 1784 violation of RFC3280) using the OpenSSL certificate creation utilities. 1785 This is done by creating a random 64 bit value for the initial serial 1786 number when a serial number file is created or when a self signed 1787 certificate is created using 'openssl req -x509'. The initial serial 1788 number file is created using 'openssl x509 -next_serial' in CA.pl 1789 rather than being initialized to 1. 1790 [Steve Henson] 1791 1792 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 1793 1794 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 1795 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 1796 [Joe Orton, Steve Henson] 1797 1798 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 1799 (CVE-2004-0112) 1800 [Joe Orton, Steve Henson] 1801 1802 *) Make it possible to have multiple active certificates with the same 1803 subject in the CA index file. This is done only if the keyword 1804 'unique_subject' is set to 'no' in the main CA section (default 1805 if 'CA_default') of the configuration file. The value is saved 1806 with the database itself in a separate index attribute file, 1807 named like the index file with '.attr' appended to the name. 1808 [Richard Levitte] 1809 1810 *) X509 verify fixes. Disable broken certificate workarounds when 1811 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 1812 keyUsage extension present. Don't accept CRLs with unhandled critical 1813 extensions: since verify currently doesn't process CRL extensions this 1814 rejects a CRL with *any* critical extensions. Add new verify error codes 1815 for these cases. 1816 [Steve Henson] 1817 1818 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 1819 A clarification of RFC2560 will require the use of OCTET STRINGs and 1820 some implementations cannot handle the current raw format. Since OpenSSL 1821 copies and compares OCSP nonces as opaque blobs without any attempt at 1822 parsing them this should not create any compatibility issues. 1823 [Steve Henson] 1824 1825 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 1826 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 1827 this HMAC (and other) operations are several times slower than OpenSSL 1828 < 0.9.7. 1829 [Steve Henson] 1830 1831 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 1832 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 1833 1834 *) Use the correct content when signing type "other". 1835 [Steve Henson] 1836 1837 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 1838 1839 *) Fix various bugs revealed by running the NISCC test suite: 1840 1841 Stop out of bounds reads in the ASN1 code when presented with 1842 invalid tags (CVE-2003-0543 and CVE-2003-0544). 1843 1844 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 1845 1846 If verify callback ignores invalid public key errors don't try to check 1847 certificate signature with the NULL public key. 1848 1849 [Steve Henson] 1850 1851 *) New -ignore_err option in ocsp application to stop the server 1852 exiting on the first error in a request. 1853 [Steve Henson] 1854 1855 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 1856 if the server requested one: as stated in TLS 1.0 and SSL 3.0 1857 specifications. 1858 [Steve Henson] 1859 1860 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 1861 extra data after the compression methods not only for TLS 1.0 1862 but also for SSL 3.0 (as required by the specification). 1863 [Bodo Moeller; problem pointed out by Matthias Loepfe] 1864 1865 *) Change X509_certificate_type() to mark the key as exported/exportable 1866 when it's 512 *bits* long, not 512 bytes. 1867 [Richard Levitte] 1868 1869 *) Change AES_cbc_encrypt() so it outputs exact multiple of 1870 blocks during encryption. 1871 [Richard Levitte] 1872 1873 *) Various fixes to base64 BIO and non blocking I/O. On write 1874 flushes were not handled properly if the BIO retried. On read 1875 data was not being buffered properly and had various logic bugs. 1876 This also affects blocking I/O when the data being decoded is a 1877 certain size. 1878 [Steve Henson] 1879 1880 *) Various S/MIME bugfixes and compatibility changes: 1881 output correct application/pkcs7 MIME type if 1882 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 1883 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 1884 of files as .eml work). Correctly handle very long lines in MIME 1885 parser. 1886 [Steve Henson] 1887 1888 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 1889 1890 *) Countermeasure against the Klima-Pokorny-Rosa extension of 1891 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 1892 a protocol version number mismatch like a decryption error 1893 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 1894 [Bodo Moeller] 1895 1896 *) Turn on RSA blinding by default in the default implementation 1897 to avoid a timing attack. Applications that don't want it can call 1898 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 1899 They would be ill-advised to do so in most cases. 1900 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 1901 1902 *) Change RSA blinding code so that it works when the PRNG is not 1903 seeded (in this case, the secret RSA exponent is abused as 1904 an unpredictable seed -- if it is not unpredictable, there 1905 is no point in blinding anyway). Make RSA blinding thread-safe 1906 by remembering the creator's thread ID in rsa->blinding and 1907 having all other threads use local one-time blinding factors 1908 (this requires more computation than sharing rsa->blinding, but 1909 avoids excessive locking; and if an RSA object is not shared 1910 between threads, blinding will still be very fast). 1911 [Bodo Moeller] 1912 1913 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 1914 ENGINE as defaults for all supported algorithms irrespective of 1915 the 'flags' parameter. 'flags' is now honoured, so applications 1916 should make sure they are passing it correctly. 1917 [Geoff Thorpe] 1918 1919 *) Target "mingw" now allows native Windows code to be generated in 1920 the Cygwin environment as well as with the MinGW compiler. 1921 [Ulf Moeller] 1922 1923 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 1924 1925 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 1926 via timing by performing a MAC computation even if incorrrect 1927 block cipher padding has been found. This is a countermeasure 1928 against active attacks where the attacker has to distinguish 1929 between bad padding and a MAC verification error. (CVE-2003-0078) 1930 1931 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 1932 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 1933 Martin Vuagnoux (EPFL, Ilion)] 1934 1935 *) Make the no-err option work as intended. The intention with no-err 1936 is not to have the whole error stack handling routines removed from 1937 libcrypto, it's only intended to remove all the function name and 1938 reason texts, thereby removing some of the footprint that may not 1939 be interesting if those errors aren't displayed anyway. 1940 1941 NOTE: it's still possible for any application or module to have it's 1942 own set of error texts inserted. The routines are there, just not 1943 used by default when no-err is given. 1944 [Richard Levitte] 1945 1946 *) Add support for FreeBSD on IA64. 1947 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 1948 1949 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 1950 Kerberos function mit_des_cbc_cksum(). Before this change, 1951 the value returned by DES_cbc_cksum() was like the one from 1952 mit_des_cbc_cksum(), except the bytes were swapped. 1953 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 1954 1955 *) Allow an application to disable the automatic SSL chain building. 1956 Before this a rather primitive chain build was always performed in 1957 ssl3_output_cert_chain(): an application had no way to send the 1958 correct chain if the automatic operation produced an incorrect result. 1959 1960 Now the chain builder is disabled if either: 1961 1962 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 1963 1964 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 1965 1966 The reasoning behind this is that an application would not want the 1967 auto chain building to take place if extra chain certificates are 1968 present and it might also want a means of sending no additional 1969 certificates (for example the chain has two certificates and the 1970 root is omitted). 1971 [Steve Henson] 1972 1973 *) Add the possibility to build without the ENGINE framework. 1974 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 1975 1976 *) Under Win32 gmtime() can return NULL: check return value in 1977 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 1978 [Steve Henson] 1979 1980 *) DSA routines: under certain error conditions uninitialized BN objects 1981 could be freed. Solution: make sure initialization is performed early 1982 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 1983 Nils Larsch <nla@trustcenter.de> via PR#459) 1984 [Lutz Jaenicke] 1985 1986 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 1987 checked on reconnect on the client side, therefore session resumption 1988 could still fail with a "ssl session id is different" error. This 1989 behaviour is masked when SSL_OP_ALL is used due to 1990 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 1991 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 1992 followup to PR #377. 1993 [Lutz Jaenicke] 1994 1995 *) IA-32 assembler support enhancements: unified ELF targets, support 1996 for SCO/Caldera platforms, fix for Cygwin shared build. 1997 [Andy Polyakov] 1998 1999 *) Add support for FreeBSD on sparc64. As a consequence, support for 2000 FreeBSD on non-x86 processors is separate from x86 processors on 2001 the config script, much like the NetBSD support. 2002 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 2003 2004 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 2005 2006 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 2007 OpenSSL 0.9.7.] 2008 2009 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 2010 code (06) was taken as the first octet of the session ID and the last 2011 octet was ignored consequently. As a result SSLv2 client side session 2012 caching could not have worked due to the session ID mismatch between 2013 client and server. 2014 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2015 PR #377. 2016 [Lutz Jaenicke] 2017 2018 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 2019 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 2020 removed entirely. 2021 [Richard Levitte] 2022 2023 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 2024 seems that in spite of existing for more than a year, many application 2025 author have done nothing to provide the necessary callbacks, which 2026 means that this particular engine will not work properly anywhere. 2027 This is a very unfortunate situation which forces us, in the name 2028 of usability, to give the hw_ncipher.c a static lock, which is part 2029 of libcrypto. 2030 NOTE: This is for the 0.9.7 series ONLY. This hack will never 2031 appear in 0.9.8 or later. We EXPECT application authors to have 2032 dealt properly with this when 0.9.8 is released (unless we actually 2033 make such changes in the libcrypto locking code that changes will 2034 have to be made anyway). 2035 [Richard Levitte] 2036 2037 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 2038 octets have been read, EOF or an error occurs. Without this change 2039 some truncated ASN1 structures will not produce an error. 2040 [Steve Henson] 2041 2042 *) Disable Heimdal support, since it hasn't been fully implemented. 2043 Still give the possibility to force the use of Heimdal, but with 2044 warnings and a request that patches get sent to openssl-dev. 2045 [Richard Levitte] 2046 2047 *) Add the VC-CE target, introduce the WINCE sysname, and add 2048 INSTALL.WCE and appropriate conditionals to make it build. 2049 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2050 2051 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 2052 cygssl-x.y.z.dll, where x, y and z are the major, minor and 2053 edit numbers of the version. 2054 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2055 2056 *) Introduce safe string copy and catenation functions 2057 (BUF_strlcpy() and BUF_strlcat()). 2058 [Ben Laurie (CHATS) and Richard Levitte] 2059 2060 *) Avoid using fixed-size buffers for one-line DNs. 2061 [Ben Laurie (CHATS)] 2062 2063 *) Add BUF_MEM_grow_clean() to avoid information leakage when 2064 resizing buffers containing secrets, and use where appropriate. 2065 [Ben Laurie (CHATS)] 2066 2067 *) Avoid using fixed size buffers for configuration file location. 2068 [Ben Laurie (CHATS)] 2069 2070 *) Avoid filename truncation for various CA files. 2071 [Ben Laurie (CHATS)] 2072 2073 *) Use sizeof in preference to magic numbers. 2074 [Ben Laurie (CHATS)] 2075 2076 *) Avoid filename truncation in cert requests. 2077 [Ben Laurie (CHATS)] 2078 2079 *) Add assertions to check for (supposedly impossible) buffer 2080 overflows. 2081 [Ben Laurie (CHATS)] 2082 2083 *) Don't cache truncated DNS entries in the local cache (this could 2084 potentially lead to a spoofing attack). 2085 [Ben Laurie (CHATS)] 2086 2087 *) Fix various buffers to be large enough for hex/decimal 2088 representations in a platform independent manner. 2089 [Ben Laurie (CHATS)] 2090 2091 *) Add CRYPTO_realloc_clean() to avoid information leakage when 2092 resizing buffers containing secrets, and use where appropriate. 2093 [Ben Laurie (CHATS)] 2094 2095 *) Add BIO_indent() to avoid much slightly worrying code to do 2096 indents. 2097 [Ben Laurie (CHATS)] 2098 2099 *) Convert sprintf()/BIO_puts() to BIO_printf(). 2100 [Ben Laurie (CHATS)] 2101 2102 *) buffer_gets() could terminate with the buffer only half 2103 full. Fixed. 2104 [Ben Laurie (CHATS)] 2105 2106 *) Add assertions to prevent user-supplied crypto functions from 2107 overflowing internal buffers by having large block sizes, etc. 2108 [Ben Laurie (CHATS)] 2109 2110 *) New OPENSSL_assert() macro (similar to assert(), but enabled 2111 unconditionally). 2112 [Ben Laurie (CHATS)] 2113 2114 *) Eliminate unused copy of key in RC4. 2115 [Ben Laurie (CHATS)] 2116 2117 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 2118 [Ben Laurie (CHATS)] 2119 2120 *) Fix off-by-one error in EGD path. 2121 [Ben Laurie (CHATS)] 2122 2123 *) If RANDFILE path is too long, ignore instead of truncating. 2124 [Ben Laurie (CHATS)] 2125 2126 *) Eliminate unused and incorrectly sized X.509 structure 2127 CBCParameter. 2128 [Ben Laurie (CHATS)] 2129 2130 *) Eliminate unused and dangerous function knumber(). 2131 [Ben Laurie (CHATS)] 2132 2133 *) Eliminate unused and dangerous structure, KSSL_ERR. 2134 [Ben Laurie (CHATS)] 2135 2136 *) Protect against overlong session ID context length in an encoded 2137 session object. Since these are local, this does not appear to be 2138 exploitable. 2139 [Ben Laurie (CHATS)] 2140 2141 *) Change from security patch (see 0.9.6e below) that did not affect 2142 the 0.9.6 release series: 2143 2144 Remote buffer overflow in SSL3 protocol - an attacker could 2145 supply an oversized master key in Kerberos-enabled versions. 2146 (CVE-2002-0657) 2147 [Ben Laurie (CHATS)] 2148 2149 *) Change the SSL kerb5 codes to match RFC 2712. 2150 [Richard Levitte] 2151 2152 *) Make -nameopt work fully for req and add -reqopt switch. 2153 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 2154 2155 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 2156 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 2157 2158 *) Make sure tests can be performed even if the corresponding algorithms 2159 have been removed entirely. This was also the last step to make 2160 OpenSSL compilable with DJGPP under all reasonable conditions. 2161 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 2162 2163 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 2164 to allow version independent disabling of normally unselected ciphers, 2165 which may be activated as a side-effect of selecting a single cipher. 2166 2167 (E.g., cipher list string "RSA" enables ciphersuites that are left 2168 out of "ALL" because they do not provide symmetric encryption. 2169 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 2170 [Lutz Jaenicke, Bodo Moeller] 2171 2172 *) Add appropriate support for separate platform-dependent build 2173 directories. The recommended way to make a platform-dependent 2174 build directory is the following (tested on Linux), maybe with 2175 some local tweaks: 2176 2177 # Place yourself outside of the OpenSSL source tree. In 2178 # this example, the environment variable OPENSSL_SOURCE 2179 # is assumed to contain the absolute OpenSSL source directory. 2180 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 2181 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 2182 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 2183 mkdir -p `dirname $F` 2184 ln -s $OPENSSL_SOURCE/$F $F 2185 done 2186 2187 To be absolutely sure not to disturb the source tree, a "make clean" 2188 is a good thing. If it isn't successfull, don't worry about it, 2189 it probably means the source directory is very clean. 2190 [Richard Levitte] 2191 2192 *) Make sure any ENGINE control commands make local copies of string 2193 pointers passed to them whenever necessary. Otherwise it is possible 2194 the caller may have overwritten (or deallocated) the original string 2195 data when a later ENGINE operation tries to use the stored values. 2196 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 2197 2198 *) Improve diagnostics in file reading and command-line digests. 2199 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 2200 2201 *) Add AES modes CFB and OFB to the object database. Correct an 2202 error in AES-CFB decryption. 2203 [Richard Levitte] 2204 2205 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 2206 allows existing EVP_CIPHER_CTX structures to be reused after 2207 calling EVP_*Final(). This behaviour is used by encryption 2208 BIOs and some applications. This has the side effect that 2209 applications must explicitly clean up cipher contexts with 2210 EVP_CIPHER_CTX_cleanup() or they will leak memory. 2211 [Steve Henson] 2212 2213 *) Check the values of dna and dnb in bn_mul_recursive before calling 2214 bn_mul_comba (a non zero value means the a or b arrays do not contain 2215 n2 elements) and fallback to bn_mul_normal if either is not zero. 2216 [Steve Henson] 2217 2218 *) Fix escaping of non-ASCII characters when using the -subj option 2219 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 2220 [Lutz Jaenicke] 2221 2222 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 2223 form for "surname", serialNumber has no short form. 2224 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 2225 therefore remove "mail" short name for "internet 7". 2226 The OID for unique identifiers in X509 certificates is 2227 x500UniqueIdentifier, not uniqueIdentifier. 2228 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 2229 [Lutz Jaenicke] 2230 2231 *) Add an "init" command to the ENGINE config module and auto initialize 2232 ENGINEs. Without any "init" command the ENGINE will be initialized 2233 after all ctrl commands have been executed on it. If init=1 the 2234 ENGINE is initailized at that point (ctrls before that point are run 2235 on the uninitialized ENGINE and after on the initialized one). If 2236 init=0 then the ENGINE will not be iniatialized at all. 2237 [Steve Henson] 2238 2239 *) Fix the 'app_verify_callback' interface so that the user-defined 2240 argument is actually passed to the callback: In the 2241 SSL_CTX_set_cert_verify_callback() prototype, the callback 2242 declaration has been changed from 2243 int (*cb)() 2244 into 2245 int (*cb)(X509_STORE_CTX *,void *); 2246 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 2247 i=s->ctx->app_verify_callback(&ctx) 2248 has been changed into 2249 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 2250 2251 To update applications using SSL_CTX_set_cert_verify_callback(), 2252 a dummy argument can be added to their callback functions. 2253 [D. K. Smetters <smetters@parc.xerox.com>] 2254 2255 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 2256 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 2257 2258 *) Add and OPENSSL_LOAD_CONF define which will cause 2259 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 2260 This allows older applications to transparently support certain 2261 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 2262 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 2263 load the config file and OPENSSL_add_all_algorithms_conf() which will 2264 always load it have also been added. 2265 [Steve Henson] 2266 2267 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 2268 Adjust NIDs and EVP layer. 2269 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2270 2271 *) Config modules support in openssl utility. 2272 2273 Most commands now load modules from the config file, 2274 though in a few (such as version) this isn't done 2275 because it couldn't be used for anything. 2276 2277 In the case of ca and req the config file used is 2278 the same as the utility itself: that is the -config 2279 command line option can be used to specify an 2280 alternative file. 2281 [Steve Henson] 2282 2283 *) Move default behaviour from OPENSSL_config(). If appname is NULL 2284 use "openssl_conf" if filename is NULL use default openssl config file. 2285 [Steve Henson] 2286 2287 *) Add an argument to OPENSSL_config() to allow the use of an alternative 2288 config section name. Add a new flag to tolerate a missing config file 2289 and move code to CONF_modules_load_file(). 2290 [Steve Henson] 2291 2292 *) Support for crypto accelerator cards from Accelerated Encryption 2293 Processing, www.aep.ie. (Use engine 'aep') 2294 The support was copied from 0.9.6c [engine] and adapted/corrected 2295 to work with the new engine framework. 2296 [AEP Inc. and Richard Levitte] 2297 2298 *) Support for SureWare crypto accelerator cards from Baltimore 2299 Technologies. (Use engine 'sureware') 2300 The support was copied from 0.9.6c [engine] and adapted 2301 to work with the new engine framework. 2302 [Richard Levitte] 2303 2304 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 2305 make the newer ENGINE framework commands for the CHIL engine work. 2306 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 2307 2308 *) Make it possible to produce shared libraries on ReliantUNIX. 2309 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 2310 2311 *) Add the configuration target debug-linux-ppro. 2312 Make 'openssl rsa' use the general key loading routines 2313 implemented in apps.c, and make those routines able to 2314 handle the key format FORMAT_NETSCAPE and the variant 2315 FORMAT_IISSGC. 2316 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2317 2318 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2319 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2320 2321 *) Add -keyform to rsautl, and document -engine. 2322 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 2323 2324 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 2325 BIO_R_NO_SUCH_FILE error code rather than the generic 2326 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 2327 [Ben Laurie] 2328 2329 *) Add new functions 2330 ERR_peek_last_error 2331 ERR_peek_last_error_line 2332 ERR_peek_last_error_line_data. 2333 These are similar to 2334 ERR_peek_error 2335 ERR_peek_error_line 2336 ERR_peek_error_line_data, 2337 but report on the latest error recorded rather than the first one 2338 still in the error queue. 2339 [Ben Laurie, Bodo Moeller] 2340 2341 *) default_algorithms option in ENGINE config module. This allows things 2342 like: 2343 default_algorithms = ALL 2344 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 2345 [Steve Henson] 2346 2347 *) Prelminary ENGINE config module. 2348 [Steve Henson] 2349 2350 *) New experimental application configuration code. 2351 [Steve Henson] 2352 2353 *) Change the AES code to follow the same name structure as all other 2354 symmetric ciphers, and behave the same way. Move everything to 2355 the directory crypto/aes, thereby obsoleting crypto/rijndael. 2356 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2357 2358 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 2359 [Ben Laurie and Theo de Raadt] 2360 2361 *) Add option to output public keys in req command. 2362 [Massimiliano Pala madwolf@openca.org] 2363 2364 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 2365 (up to about 10% better than before for P-192 and P-224). 2366 [Bodo Moeller] 2367 2368 *) New functions/macros 2369 2370 SSL_CTX_set_msg_callback(ctx, cb) 2371 SSL_CTX_set_msg_callback_arg(ctx, arg) 2372 SSL_set_msg_callback(ssl, cb) 2373 SSL_set_msg_callback_arg(ssl, arg) 2374 2375 to request calling a callback function 2376 2377 void cb(int write_p, int version, int content_type, 2378 const void *buf, size_t len, SSL *ssl, void *arg) 2379 2380 whenever a protocol message has been completely received 2381 (write_p == 0) or sent (write_p == 1). Here 'version' is the 2382 protocol version according to which the SSL library interprets 2383 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 2384 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 2385 the content type as defined in the SSL 3.0/TLS 1.0 protocol 2386 specification (change_cipher_spec(20), alert(21), handshake(22)). 2387 'buf' and 'len' point to the actual message, 'ssl' to the 2388 SSL object, and 'arg' is the application-defined value set by 2389 SSL[_CTX]_set_msg_callback_arg(). 2390 2391 'openssl s_client' and 'openssl s_server' have new '-msg' options 2392 to enable a callback that displays all protocol messages. 2393 [Bodo Moeller] 2394 2395 *) Change the shared library support so shared libraries are built as 2396 soon as the corresponding static library is finished, and thereby get 2397 openssl and the test programs linked against the shared library. 2398 This still only happens when the keyword "shard" has been given to 2399 the configuration scripts. 2400 2401 NOTE: shared library support is still an experimental thing, and 2402 backward binary compatibility is still not guaranteed. 2403 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 2404 2405 *) Add support for Subject Information Access extension. 2406 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2407 2408 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 2409 additional bytes when new memory had to be allocated, not just 2410 when reusing an existing buffer. 2411 [Bodo Moeller] 2412 2413 *) New command line and configuration option 'utf8' for the req command. 2414 This allows field values to be specified as UTF8 strings. 2415 [Steve Henson] 2416 2417 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 2418 runs for the former and machine-readable output for the latter. 2419 [Ben Laurie] 2420 2421 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 2422 of the e-mail address in the DN (i.e., it will go into a certificate 2423 extension only). The new configuration file option 'email_in_dn = no' 2424 has the same effect. 2425 [Massimiliano Pala madwolf@openca.org] 2426 2427 *) Change all functions with names starting with des_ to be starting 2428 with DES_ instead. Add wrappers that are compatible with libdes, 2429 but are named _ossl_old_des_*. Finally, add macros that map the 2430 des_* symbols to the corresponding _ossl_old_des_* if libdes 2431 compatibility is desired. If OpenSSL 0.9.6c compatibility is 2432 desired, the des_* symbols will be mapped to DES_*, with one 2433 exception. 2434 2435 Since we provide two compatibility mappings, the user needs to 2436 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 2437 compatibility is desired. The default (i.e., when that macro 2438 isn't defined) is OpenSSL 0.9.6c compatibility. 2439 2440 There are also macros that enable and disable the support of old 2441 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 2442 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 2443 are defined, the default will apply: to support the old des routines. 2444 2445 In either case, one must include openssl/des.h to get the correct 2446 definitions. Do not try to just include openssl/des_old.h, that 2447 won't work. 2448 2449 NOTE: This is a major break of an old API into a new one. Software 2450 authors are encouraged to switch to the DES_ style functions. Some 2451 time in the future, des_old.h and the libdes compatibility functions 2452 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 2453 default), and then completely removed. 2454 [Richard Levitte] 2455 2456 *) Test for certificates which contain unsupported critical extensions. 2457 If such a certificate is found during a verify operation it is 2458 rejected by default: this behaviour can be overridden by either 2459 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 2460 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 2461 X509_supported_extension() has also been added which returns 1 if a 2462 particular extension is supported. 2463 [Steve Henson] 2464 2465 *) Modify the behaviour of EVP cipher functions in similar way to digests 2466 to retain compatibility with existing code. 2467 [Steve Henson] 2468 2469 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 2470 compatibility with existing code. In particular the 'ctx' parameter does 2471 not have to be to be initialized before the call to EVP_DigestInit() and 2472 it is tidied up after a call to EVP_DigestFinal(). New function 2473 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 2474 EVP_MD_CTX_copy() changed to not require the destination to be 2475 initialized valid and new function EVP_MD_CTX_copy_ex() added which 2476 requires the destination to be valid. 2477 2478 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 2479 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 2480 [Steve Henson] 2481 2482 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 2483 so that complete 'Handshake' protocol structures are kept in memory 2484 instead of overwriting 'msg_type' and 'length' with 'body' data. 2485 [Bodo Moeller] 2486 2487 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 2488 [Massimo Santin via Richard Levitte] 2489 2490 *) Major restructuring to the underlying ENGINE code. This includes 2491 reduction of linker bloat, separation of pure "ENGINE" manipulation 2492 (initialisation, etc) from functionality dealing with implementations 2493 of specific crypto iterfaces. This change also introduces integrated 2494 support for symmetric ciphers and digest implementations - so ENGINEs 2495 can now accelerate these by providing EVP_CIPHER and EVP_MD 2496 implementations of their own. This is detailed in crypto/engine/README 2497 as it couldn't be adequately described here. However, there are a few 2498 API changes worth noting - some RSA, DSA, DH, and RAND functions that 2499 were changed in the original introduction of ENGINE code have now 2500 reverted back - the hooking from this code to ENGINE is now a good 2501 deal more passive and at run-time, operations deal directly with 2502 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 2503 dereferencing through an ENGINE pointer any more. Also, the ENGINE 2504 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 2505 they were not being used by the framework as there is no concept of a 2506 BIGNUM_METHOD and they could not be generalised to the new 2507 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 2508 ENGINE_cpy() has been removed as it cannot be consistently defined in 2509 the new code. 2510 [Geoff Thorpe] 2511 2512 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 2513 [Steve Henson] 2514 2515 *) Change mkdef.pl to sort symbols that get the same entry number, 2516 and make sure the automatically generated functions ERR_load_* 2517 become part of libeay.num as well. 2518 [Richard Levitte] 2519 2520 *) New function SSL_renegotiate_pending(). This returns true once 2521 renegotiation has been requested (either SSL_renegotiate() call 2522 or HelloRequest/ClientHello receveived from the peer) and becomes 2523 false once a handshake has been completed. 2524 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 2525 sends a HelloRequest, but does not ensure that a handshake takes 2526 place. SSL_renegotiate_pending() is useful for checking if the 2527 client has followed the request.) 2528 [Bodo Moeller] 2529 2530 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 2531 By default, clients may request session resumption even during 2532 renegotiation (if session ID contexts permit); with this option, 2533 session resumption is possible only in the first handshake. 2534 2535 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 2536 more bits available for options that should not be part of 2537 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 2538 [Bodo Moeller] 2539 2540 *) Add some demos for certificate and certificate request creation. 2541 [Steve Henson] 2542 2543 *) Make maximum certificate chain size accepted from the peer application 2544 settable (SSL*_get/set_max_cert_list()), as proposed by 2545 "Douglas E. Engert" <deengert@anl.gov>. 2546 [Lutz Jaenicke] 2547 2548 *) Add support for shared libraries for Unixware-7 2549 (Boyd Lynn Gerber <gerberb@zenez.com>). 2550 [Lutz Jaenicke] 2551 2552 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 2553 be done prior to destruction. Use this to unload error strings from 2554 ENGINEs that load their own error strings. NB: This adds two new API 2555 functions to "get" and "set" this destroy handler in an ENGINE. 2556 [Geoff Thorpe] 2557 2558 *) Alter all existing ENGINE implementations (except "openssl" and 2559 "openbsd") to dynamically instantiate their own error strings. This 2560 makes them more flexible to be built both as statically-linked ENGINEs 2561 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 2562 Also, add stub code to each that makes building them as self-contained 2563 shared-libraries easier (see README.ENGINE). 2564 [Geoff Thorpe] 2565 2566 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 2567 implementations into applications that are completely implemented in 2568 self-contained shared-libraries. The "dynamic" ENGINE exposes control 2569 commands that can be used to configure what shared-library to load and 2570 to control aspects of the way it is handled. Also, made an update to 2571 the README.ENGINE file that brings its information up-to-date and 2572 provides some information and instructions on the "dynamic" ENGINE 2573 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 2574 [Geoff Thorpe] 2575 2576 *) Make it possible to unload ranges of ERR strings with a new 2577 "ERR_unload_strings" function. 2578 [Geoff Thorpe] 2579 2580 *) Add a copy() function to EVP_MD. 2581 [Ben Laurie] 2582 2583 *) Make EVP_MD routines take a context pointer instead of just the 2584 md_data void pointer. 2585 [Ben Laurie] 2586 2587 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 2588 that the digest can only process a single chunk of data 2589 (typically because it is provided by a piece of 2590 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 2591 is only going to provide a single chunk of data, and hence the 2592 framework needn't accumulate the data for oneshot drivers. 2593 [Ben Laurie] 2594 2595 *) As with "ERR", make it possible to replace the underlying "ex_data" 2596 functions. This change also alters the storage and management of global 2597 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 2598 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 2599 index counters. The API functions that use this state have been changed 2600 to take a "class_index" rather than pointers to the class's local STACK 2601 and counter, and there is now an API function to dynamically create new 2602 classes. This centralisation allows us to (a) plug a lot of the 2603 thread-safety problems that existed, and (b) makes it possible to clean 2604 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 2605 such data would previously have always leaked in application code and 2606 workarounds were in place to make the memory debugging turn a blind eye 2607 to it. Application code that doesn't use this new function will still 2608 leak as before, but their memory debugging output will announce it now 2609 rather than letting it slide. 2610 2611 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 2612 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 2613 has a return value to indicate success or failure. 2614 [Geoff Thorpe] 2615 2616 *) Make it possible to replace the underlying "ERR" functions such that the 2617 global state (2 LHASH tables and 2 locks) is only used by the "default" 2618 implementation. This change also adds two functions to "get" and "set" 2619 the implementation prior to it being automatically set the first time 2620 any other ERR function takes place. Ie. an application can call "get", 2621 pass the return value to a module it has just loaded, and that module 2622 can call its own "set" function using that value. This means the 2623 module's "ERR" operations will use (and modify) the error state in the 2624 application and not in its own statically linked copy of OpenSSL code. 2625 [Geoff Thorpe] 2626 2627 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 2628 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 2629 the operation, and provides a more encapsulated way for external code 2630 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 2631 to use these functions rather than manually incrementing the counts. 2632 2633 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 2634 [Geoff Thorpe] 2635 2636 *) Add EVP test program. 2637 [Ben Laurie] 2638 2639 *) Add symmetric cipher support to ENGINE. Expect the API to change! 2640 [Ben Laurie] 2641 2642 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 2643 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 2644 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 2645 These allow a CRL to be built without having to access X509_CRL fields 2646 directly. Modify 'ca' application to use new functions. 2647 [Steve Henson] 2648 2649 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 2650 bug workarounds. Rollback attack detection is a security feature. 2651 The problem will only arise on OpenSSL servers when TLSv1 is not 2652 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 2653 Software authors not wanting to support TLSv1 will have special reasons 2654 for their choice and can explicitly enable this option. 2655 [Bodo Moeller, Lutz Jaenicke] 2656 2657 *) Rationalise EVP so it can be extended: don't include a union of 2658 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 2659 (similar to those existing for EVP_CIPHER_CTX). 2660 Usage example: 2661 2662 EVP_MD_CTX md; 2663 2664 EVP_MD_CTX_init(&md); /* new function call */ 2665 EVP_DigestInit(&md, EVP_sha1()); 2666 EVP_DigestUpdate(&md, in, len); 2667 EVP_DigestFinal(&md, out, NULL); 2668 EVP_MD_CTX_cleanup(&md); /* new function call */ 2669 2670 [Ben Laurie] 2671 2672 *) Make DES key schedule conform to the usual scheme, as well as 2673 correcting its structure. This means that calls to DES functions 2674 now have to pass a pointer to a des_key_schedule instead of a 2675 plain des_key_schedule (which was actually always a pointer 2676 anyway): E.g., 2677 2678 des_key_schedule ks; 2679 2680 des_set_key_checked(..., &ks); 2681 des_ncbc_encrypt(..., &ks, ...); 2682 2683 (Note that a later change renames 'des_...' into 'DES_...'.) 2684 [Ben Laurie] 2685 2686 *) Initial reduction of linker bloat: the use of some functions, such as 2687 PEM causes large amounts of unused functions to be linked in due to 2688 poor organisation. For example pem_all.c contains every PEM function 2689 which has a knock on effect of linking in large amounts of (unused) 2690 ASN1 code. Grouping together similar functions and splitting unrelated 2691 functions prevents this. 2692 [Steve Henson] 2693 2694 *) Cleanup of EVP macros. 2695 [Ben Laurie] 2696 2697 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 2698 correct _ecb suffix. 2699 [Ben Laurie] 2700 2701 *) Add initial OCSP responder support to ocsp application. The 2702 revocation information is handled using the text based index 2703 use by the ca application. The responder can either handle 2704 requests generated internally, supplied in files (for example 2705 via a CGI script) or using an internal minimal server. 2706 [Steve Henson] 2707 2708 *) Add configuration choices to get zlib compression for TLS. 2709 [Richard Levitte] 2710 2711 *) Changes to Kerberos SSL for RFC 2712 compliance: 2712 1. Implemented real KerberosWrapper, instead of just using 2713 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 2714 2. Implemented optional authenticator field of KerberosWrapper. 2715 2716 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 2717 and authenticator structs; see crypto/krb5/. 2718 2719 Generalized Kerberos calls to support multiple Kerberos libraries. 2720 [Vern Staats <staatsvr@asc.hpc.mil>, 2721 Jeffrey Altman <jaltman@columbia.edu> 2722 via Richard Levitte] 2723 2724 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 2725 already does with RSA. testdsa.h now has 'priv_key/pub_key' 2726 values for each of the key sizes rather than having just 2727 parameters (and 'speed' generating keys each time). 2728 [Geoff Thorpe] 2729 2730 *) Speed up EVP routines. 2731 Before: 2732encrypt 2733type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 2734des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 2735des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 2736des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 2737decrypt 2738des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 2739des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 2740des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 2741 After: 2742encrypt 2743des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 2744decrypt 2745des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 2746 [Ben Laurie] 2747 2748 *) Added the OS2-EMX target. 2749 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 2750 2751 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 2752 to support NCONF routines in extension code. New function CONF_set_nconf() 2753 to allow functions which take an NCONF to also handle the old LHASH 2754 structure: this means that the old CONF compatible routines can be 2755 retained (in particular wrt extensions) without having to duplicate the 2756 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 2757 [Steve Henson] 2758 2759 *) Enhance the general user interface with mechanisms for inner control 2760 and with possibilities to have yes/no kind of prompts. 2761 [Richard Levitte] 2762 2763 *) Change all calls to low level digest routines in the library and 2764 applications to use EVP. Add missing calls to HMAC_cleanup() and 2765 don't assume HMAC_CTX can be copied using memcpy(). 2766 [Verdon Walker <VWalker@novell.com>, Steve Henson] 2767 2768 *) Add the possibility to control engines through control names but with 2769 arbitrary arguments instead of just a string. 2770 Change the key loaders to take a UI_METHOD instead of a callback 2771 function pointer. NOTE: this breaks binary compatibility with earlier 2772 versions of OpenSSL [engine]. 2773 Adapt the nCipher code for these new conditions and add a card insertion 2774 callback. 2775 [Richard Levitte] 2776 2777 *) Enhance the general user interface with mechanisms to better support 2778 dialog box interfaces, application-defined prompts, the possibility 2779 to use defaults (for example default passwords from somewhere else) 2780 and interrupts/cancellations. 2781 [Richard Levitte] 2782 2783 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 2784 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 2785 [Steve Henson] 2786 2787 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 2788 tidy up some unnecessarily weird code in 'sk_new()'). 2789 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 2790 2791 *) Change the key loading routines for ENGINEs to use the same kind 2792 callback (pem_password_cb) as all other routines that need this 2793 kind of callback. 2794 [Richard Levitte] 2795 2796 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 2797 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 2798 than this minimum value is recommended. 2799 [Lutz Jaenicke] 2800 2801 *) New random seeder for OpenVMS, using the system process statistics 2802 that are easily reachable. 2803 [Richard Levitte] 2804 2805 *) Windows apparently can't transparently handle global 2806 variables defined in DLLs. Initialisations such as: 2807 2808 const ASN1_ITEM *it = &ASN1_INTEGER_it; 2809 2810 wont compile. This is used by the any applications that need to 2811 declare their own ASN1 modules. This was fixed by adding the option 2812 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 2813 needed for static libraries under Win32. 2814 [Steve Henson] 2815 2816 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 2817 setting of purpose and trust fields. New X509_STORE trust and 2818 purpose functions and tidy up setting in other SSL functions. 2819 [Steve Henson] 2820 2821 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 2822 structure. These are inherited by X509_STORE_CTX when it is 2823 initialised. This allows various defaults to be set in the 2824 X509_STORE structure (such as flags for CRL checking and custom 2825 purpose or trust settings) for functions which only use X509_STORE_CTX 2826 internally such as S/MIME. 2827 2828 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 2829 trust settings if they are not set in X509_STORE. This allows X509_STORE 2830 purposes and trust (in S/MIME for example) to override any set by default. 2831 2832 Add command line options for CRL checking to smime, s_client and s_server 2833 applications. 2834 [Steve Henson] 2835 2836 *) Initial CRL based revocation checking. If the CRL checking flag(s) 2837 are set then the CRL is looked up in the X509_STORE structure and 2838 its validity and signature checked, then if the certificate is found 2839 in the CRL the verify fails with a revoked error. 2840 2841 Various new CRL related callbacks added to X509_STORE_CTX structure. 2842 2843 Command line options added to 'verify' application to support this. 2844 2845 This needs some additional work, such as being able to handle multiple 2846 CRLs with different times, extension based lookup (rather than just 2847 by subject name) and ultimately more complete V2 CRL extension 2848 handling. 2849 [Steve Henson] 2850 2851 *) Add a general user interface API (crypto/ui/). This is designed 2852 to replace things like des_read_password and friends (backward 2853 compatibility functions using this new API are provided). 2854 The purpose is to remove prompting functions from the DES code 2855 section as well as provide for prompting through dialog boxes in 2856 a window system and the like. 2857 [Richard Levitte] 2858 2859 *) Add "ex_data" support to ENGINE so implementations can add state at a 2860 per-structure level rather than having to store it globally. 2861 [Geoff] 2862 2863 *) Make it possible for ENGINE structures to be copied when retrieved by 2864 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 2865 This causes the "original" ENGINE structure to act like a template, 2866 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 2867 operational state can be localised to each ENGINE structure, despite the 2868 fact they all share the same "methods". New ENGINE structures returned in 2869 this case have no functional references and the return value is the single 2870 structural reference. This matches the single structural reference returned 2871 by ENGINE_by_id() normally, when it is incremented on the pre-existing 2872 ENGINE structure. 2873 [Geoff] 2874 2875 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 2876 needs to match any other type at all we need to manually clear the 2877 tag cache. 2878 [Steve Henson] 2879 2880 *) Changes to the "openssl engine" utility to include; 2881 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 2882 about an ENGINE's available control commands. 2883 - executing control commands from command line arguments using the 2884 '-pre' and '-post' switches. '-post' is only used if '-t' is 2885 specified and the ENGINE is successfully initialised. The syntax for 2886 the individual commands are colon-separated, for example; 2887 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 2888 [Geoff] 2889 2890 *) New dynamic control command support for ENGINEs. ENGINEs can now 2891 declare their own commands (numbers), names (strings), descriptions, 2892 and input types for run-time discovery by calling applications. A 2893 subset of these commands are implicitly classed as "executable" 2894 depending on their input type, and only these can be invoked through 2895 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 2896 can be based on user input, config files, etc). The distinction is 2897 that "executable" commands cannot return anything other than a boolean 2898 result and can only support numeric or string input, whereas some 2899 discoverable commands may only be for direct use through 2900 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 2901 pointers, or other custom uses. The "executable" commands are to 2902 support parameterisations of ENGINE behaviour that can be 2903 unambiguously defined by ENGINEs and used consistently across any 2904 OpenSSL-based application. Commands have been added to all the 2905 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 2906 control over shared-library paths without source code alterations. 2907 [Geoff] 2908 2909 *) Changed all ENGINE implementations to dynamically allocate their 2910 ENGINEs rather than declaring them statically. Apart from this being 2911 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 2912 this also allows the implementations to compile without using the 2913 internal engine_int.h header. 2914 [Geoff] 2915 2916 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 2917 'const' value. Any code that should be able to modify a RAND_METHOD 2918 should already have non-const pointers to it (ie. they should only 2919 modify their own ones). 2920 [Geoff] 2921 2922 *) Made a variety of little tweaks to the ENGINE code. 2923 - "atalla" and "ubsec" string definitions were moved from header files 2924 to C code. "nuron" string definitions were placed in variables 2925 rather than hard-coded - allowing parameterisation of these values 2926 later on via ctrl() commands. 2927 - Removed unused "#if 0"'d code. 2928 - Fixed engine list iteration code so it uses ENGINE_free() to release 2929 structural references. 2930 - Constified the RAND_METHOD element of ENGINE structures. 2931 - Constified various get/set functions as appropriate and added 2932 missing functions (including a catch-all ENGINE_cpy that duplicates 2933 all ENGINE values onto a new ENGINE except reference counts/state). 2934 - Removed NULL parameter checks in get/set functions. Setting a method 2935 or function to NULL is a way of cancelling out a previously set 2936 value. Passing a NULL ENGINE parameter is just plain stupid anyway 2937 and doesn't justify the extra error symbols and code. 2938 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 2939 flags from engine_int.h to engine.h. 2940 - Changed prototypes for ENGINE handler functions (init(), finish(), 2941 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 2942 [Geoff] 2943 2944 *) Implement binary inversion algorithm for BN_mod_inverse in addition 2945 to the algorithm using long division. The binary algorithm can be 2946 used only if the modulus is odd. On 32-bit systems, it is faster 2947 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 2948 roughly 5-15% for 256-bit moduli), so we use it only for moduli 2949 up to 450 bits. In 64-bit environments, the binary algorithm 2950 appears to be advantageous for much longer moduli; here we use it 2951 for moduli up to 2048 bits. 2952 [Bodo Moeller] 2953 2954 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 2955 could not support the combine flag in choice fields. 2956 [Steve Henson] 2957 2958 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 2959 extensions from a certificate request to the certificate. 2960 [Steve Henson] 2961 2962 *) Allow multiple 'certopt' and 'nameopt' options to be separated 2963 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 2964 file: this allows the display of the certificate about to be 2965 signed to be customised, to allow certain fields to be included 2966 or excluded and extension details. The old system didn't display 2967 multicharacter strings properly, omitted fields not in the policy 2968 and couldn't display additional details such as extensions. 2969 [Steve Henson] 2970 2971 *) Function EC_POINTs_mul for multiple scalar multiplication 2972 of an arbitrary number of elliptic curve points 2973 \sum scalars[i]*points[i], 2974 optionally including the generator defined for the EC_GROUP: 2975 scalar*generator + \sum scalars[i]*points[i]. 2976 2977 EC_POINT_mul is a simple wrapper function for the typical case 2978 that the point list has just one item (besides the optional 2979 generator). 2980 [Bodo Moeller] 2981 2982 *) First EC_METHODs for curves over GF(p): 2983 2984 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 2985 operations and provides various method functions that can also 2986 operate with faster implementations of modular arithmetic. 2987 2988 EC_GFp_mont_method() reuses most functions that are part of 2989 EC_GFp_simple_method, but uses Montgomery arithmetic. 2990 2991 [Bodo Moeller; point addition and point doubling 2992 implementation directly derived from source code provided by 2993 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 2994 2995 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 2996 crypto/ec/ec_lib.c): 2997 2998 Curves are EC_GROUP objects (with an optional group generator) 2999 based on EC_METHODs that are built into the library. 3000 3001 Points are EC_POINT objects based on EC_GROUP objects. 3002 3003 Most of the framework would be able to handle curves over arbitrary 3004 finite fields, but as there are no obvious types for fields other 3005 than GF(p), some functions are limited to that for now. 3006 [Bodo Moeller] 3007 3008 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 3009 that the file contains a complete HTTP response. 3010 [Richard Levitte] 3011 3012 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 3013 change the def and num file printf format specifier from "%-40sXXX" 3014 to "%-39s XXX". The latter will always guarantee a space after the 3015 field while the former will cause them to run together if the field 3016 is 40 of more characters long. 3017 [Steve Henson] 3018 3019 *) Constify the cipher and digest 'method' functions and structures 3020 and modify related functions to take constant EVP_MD and EVP_CIPHER 3021 pointers. 3022 [Steve Henson] 3023 3024 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 3025 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 3026 [Bodo Moeller] 3027 3028 *) Modify EVP_Digest*() routines so they now return values. Although the 3029 internal software routines can never fail additional hardware versions 3030 might. 3031 [Steve Henson] 3032 3033 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 3034 3035 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 3036 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 3037 3038 ASN1 error codes 3039 ERR_R_NESTED_ASN1_ERROR 3040 ... 3041 ERR_R_MISSING_ASN1_EOS 3042 were 4 .. 9, conflicting with 3043 ERR_LIB_RSA (= ERR_R_RSA_LIB) 3044 ... 3045 ERR_LIB_PEM (= ERR_R_PEM_LIB). 3046 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 3047 3048 Add new error code 'ERR_R_INTERNAL_ERROR'. 3049 [Bodo Moeller] 3050 3051 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 3052 suffices. 3053 [Bodo Moeller] 3054 3055 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 3056 sets the subject name for a new request or supersedes the 3057 subject name in a given request. Formats that can be parsed are 3058 'CN=Some Name, OU=myOU, C=IT' 3059 and 3060 'CN=Some Name/OU=myOU/C=IT'. 3061 3062 Add options '-batch' and '-verbose' to 'openssl req'. 3063 [Massimiliano Pala <madwolf@hackmasters.net>] 3064 3065 *) Introduce the possibility to access global variables through 3066 functions on platform were that's the best way to handle exporting 3067 global variables in shared libraries. To enable this functionality, 3068 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 3069 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 3070 is normally done by Configure or something similar). 3071 3072 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 3073 in the source file (foo.c) like this: 3074 3075 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 3076 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 3077 3078 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 3079 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 3080 3081 OPENSSL_DECLARE_GLOBAL(int,foo); 3082 #define foo OPENSSL_GLOBAL_REF(foo) 3083 OPENSSL_DECLARE_GLOBAL(double,bar); 3084 #define bar OPENSSL_GLOBAL_REF(bar) 3085 3086 The #defines are very important, and therefore so is including the 3087 header file everywhere where the defined globals are used. 3088 3089 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 3090 of ASN.1 items, but that structure is a bit different. 3091 3092 The largest change is in util/mkdef.pl which has been enhanced with 3093 better and easier to understand logic to choose which symbols should 3094 go into the Windows .def files as well as a number of fixes and code 3095 cleanup (among others, algorithm keywords are now sorted 3096 lexicographically to avoid constant rewrites). 3097 [Richard Levitte] 3098 3099 *) In BN_div() keep a copy of the sign of 'num' before writing the 3100 result to 'rm' because if rm==num the value will be overwritten 3101 and produce the wrong result if 'num' is negative: this caused 3102 problems with BN_mod() and BN_nnmod(). 3103 [Steve Henson] 3104 3105 *) Function OCSP_request_verify(). This checks the signature on an 3106 OCSP request and verifies the signer certificate. The signer 3107 certificate is just checked for a generic purpose and OCSP request 3108 trust settings. 3109 [Steve Henson] 3110 3111 *) Add OCSP_check_validity() function to check the validity of OCSP 3112 responses. OCSP responses are prepared in real time and may only 3113 be a few seconds old. Simply checking that the current time lies 3114 between thisUpdate and nextUpdate max reject otherwise valid responses 3115 caused by either OCSP responder or client clock inaccuracy. Instead 3116 we allow thisUpdate and nextUpdate to fall within a certain period of 3117 the current time. The age of the response can also optionally be 3118 checked. Two new options -validity_period and -status_age added to 3119 ocsp utility. 3120 [Steve Henson] 3121 3122 *) If signature or public key algorithm is unrecognized print out its 3123 OID rather that just UNKNOWN. 3124 [Steve Henson] 3125 3126 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 3127 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 3128 ID to be generated from the issuer certificate alone which can then be 3129 passed to OCSP_id_issuer_cmp(). 3130 [Steve Henson] 3131 3132 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 3133 ASN1 modules to export functions returning ASN1_ITEM pointers 3134 instead of the ASN1_ITEM structures themselves. This adds several 3135 new macros which allow the underlying ASN1 function/structure to 3136 be accessed transparently. As a result code should not use ASN1_ITEM 3137 references directly (such as &X509_it) but instead use the relevant 3138 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 3139 use of the new ASN1 code on platforms where exporting structures 3140 is problematical (for example in shared libraries) but exporting 3141 functions returning pointers to structures is not. 3142 [Steve Henson] 3143 3144 *) Add support for overriding the generation of SSL/TLS session IDs. 3145 These callbacks can be registered either in an SSL_CTX or per SSL. 3146 The purpose of this is to allow applications to control, if they wish, 3147 the arbitrary values chosen for use as session IDs, particularly as it 3148 can be useful for session caching in multiple-server environments. A 3149 command-line switch for testing this (and any client code that wishes 3150 to use such a feature) has been added to "s_server". 3151 [Geoff Thorpe, Lutz Jaenicke] 3152 3153 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 3154 of the form '#if defined(...) || defined(...) || ...' and 3155 '#if !defined(...) && !defined(...) && ...'. This also avoids 3156 the growing number of special cases it was previously handling. 3157 [Richard Levitte] 3158 3159 *) Make all configuration macros available for application by making 3160 sure they are available in opensslconf.h, by giving them names starting 3161 with "OPENSSL_" to avoid conflicts with other packages and by making 3162 sure e_os2.h will cover all platform-specific cases together with 3163 opensslconf.h. 3164 Additionally, it is now possible to define configuration/platform- 3165 specific names (called "system identities"). In the C code, these 3166 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 3167 macro with the name beginning with "OPENSSL_SYS_", which is determined 3168 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 3169 what is available. 3170 [Richard Levitte] 3171 3172 *) New option -set_serial to 'req' and 'x509' this allows the serial 3173 number to use to be specified on the command line. Previously self 3174 signed certificates were hard coded with serial number 0 and the 3175 CA options of 'x509' had to use a serial number in a file which was 3176 auto incremented. 3177 [Steve Henson] 3178 3179 *) New options to 'ca' utility to support V2 CRL entry extensions. 3180 Currently CRL reason, invalidity date and hold instruction are 3181 supported. Add new CRL extensions to V3 code and some new objects. 3182 [Steve Henson] 3183 3184 *) New function EVP_CIPHER_CTX_set_padding() this is used to 3185 disable standard block padding (aka PKCS#5 padding) in the EVP 3186 API, which was previously mandatory. This means that the data is 3187 not padded in any way and so the total length much be a multiple 3188 of the block size, otherwise an error occurs. 3189 [Steve Henson] 3190 3191 *) Initial (incomplete) OCSP SSL support. 3192 [Steve Henson] 3193 3194 *) New function OCSP_parse_url(). This splits up a URL into its host, 3195 port and path components: primarily to parse OCSP URLs. New -url 3196 option to ocsp utility. 3197 [Steve Henson] 3198 3199 *) New nonce behavior. The return value of OCSP_check_nonce() now 3200 reflects the various checks performed. Applications can decide 3201 whether to tolerate certain situations such as an absent nonce 3202 in a response when one was present in a request: the ocsp application 3203 just prints out a warning. New function OCSP_add1_basic_nonce() 3204 this is to allow responders to include a nonce in a response even if 3205 the request is nonce-less. 3206 [Steve Henson] 3207 3208 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 3209 skipped when using openssl x509 multiple times on a single input file, 3210 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 3211 [Bodo Moeller] 3212 3213 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 3214 set string type: to handle setting ASN1_TIME structures. Fix ca 3215 utility to correctly initialize revocation date of CRLs. 3216 [Steve Henson] 3217 3218 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 3219 the clients preferred ciphersuites and rather use its own preferences. 3220 Should help to work around M$ SGC (Server Gated Cryptography) bug in 3221 Internet Explorer by ensuring unchanged hash method during stepup. 3222 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 3223 [Lutz Jaenicke] 3224 3225 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 3226 to aes and add a new 'exist' option to print out symbols that don't 3227 appear to exist. 3228 [Steve Henson] 3229 3230 *) Additional options to ocsp utility to allow flags to be set and 3231 additional certificates supplied. 3232 [Steve Henson] 3233 3234 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 3235 OCSP client a number of certificate to only verify the response 3236 signature against. 3237 [Richard Levitte] 3238 3239 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 3240 handle the new API. Currently only ECB, CBC modes supported. Add new 3241 AES OIDs. 3242 3243 Add TLS AES ciphersuites as described in RFC3268, "Advanced 3244 Encryption Standard (AES) Ciphersuites for Transport Layer 3245 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 3246 not enabled by default and were not part of the "ALL" ciphersuite 3247 alias because they were not yet official; they could be 3248 explicitly requested by specifying the "AESdraft" ciphersuite 3249 group alias. In the final release of OpenSSL 0.9.7, the group 3250 alias is called "AES" and is part of "ALL".) 3251 [Ben Laurie, Steve Henson, Bodo Moeller] 3252 3253 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 3254 request to response. 3255 [Steve Henson] 3256 3257 *) Functions for OCSP responders. OCSP_request_onereq_count(), 3258 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 3259 extract information from a certificate request. OCSP_response_create() 3260 creates a response and optionally adds a basic response structure. 3261 OCSP_basic_add1_status() adds a complete single response to a basic 3262 response and returns the OCSP_SINGLERESP structure just added (to allow 3263 extensions to be included for example). OCSP_basic_add1_cert() adds a 3264 certificate to a basic response and OCSP_basic_sign() signs a basic 3265 response with various flags. New helper functions ASN1_TIME_check() 3266 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 3267 (converts ASN1_TIME to GeneralizedTime). 3268 [Steve Henson] 3269 3270 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 3271 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 3272 structure from a certificate. X509_pubkey_digest() digests the public_key 3273 contents: this is used in various key identifiers. 3274 [Steve Henson] 3275 3276 *) Make sk_sort() tolerate a NULL argument. 3277 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 3278 3279 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 3280 passed by the function are trusted implicitly. If any of them signed the 3281 response then it is assumed to be valid and is not verified. 3282 [Steve Henson] 3283 3284 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 3285 to data. This was previously part of the PKCS7 ASN1 code. This 3286 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 3287 [Steve Henson, reported by Kenneth R. Robinette 3288 <support@securenetterm.com>] 3289 3290 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 3291 routines: without these tracing memory leaks is very painful. 3292 Fix leaks in PKCS12 and PKCS7 routines. 3293 [Steve Henson] 3294 3295 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 3296 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 3297 effectively meant GeneralizedTime would never be used. Now it 3298 is initialised to -1 but X509_time_adj() now has to check the value 3299 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 3300 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 3301 [Steve Henson, reported by Kenneth R. Robinette 3302 <support@securenetterm.com>] 3303 3304 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 3305 result in a zero length in the ASN1_INTEGER structure which was 3306 not consistent with the structure when d2i_ASN1_INTEGER() was used 3307 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 3308 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 3309 where it did not print out a minus for negative ASN1_INTEGER. 3310 [Steve Henson] 3311 3312 *) Add summary printout to ocsp utility. The various functions which 3313 convert status values to strings have been renamed to: 3314 OCSP_response_status_str(), OCSP_cert_status_str() and 3315 OCSP_crl_reason_str() and are no longer static. New options 3316 to verify nonce values and to disable verification. OCSP response 3317 printout format cleaned up. 3318 [Steve Henson] 3319 3320 *) Add additional OCSP certificate checks. These are those specified 3321 in RFC2560. This consists of two separate checks: the CA of the 3322 certificate being checked must either be the OCSP signer certificate 3323 or the issuer of the OCSP signer certificate. In the latter case the 3324 OCSP signer certificate must contain the OCSP signing extended key 3325 usage. This check is performed by attempting to match the OCSP 3326 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 3327 in the OCSP_CERTID structures of the response. 3328 [Steve Henson] 3329 3330 *) Initial OCSP certificate verification added to OCSP_basic_verify() 3331 and related routines. This uses the standard OpenSSL certificate 3332 verify routines to perform initial checks (just CA validity) and 3333 to obtain the certificate chain. Then additional checks will be 3334 performed on the chain. Currently the root CA is checked to see 3335 if it is explicitly trusted for OCSP signing. This is used to set 3336 a root CA as a global signing root: that is any certificate that 3337 chains to that CA is an acceptable OCSP signing certificate. 3338 [Steve Henson] 3339 3340 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 3341 extensions from a separate configuration file. 3342 As when reading extensions from the main configuration file, 3343 the '-extensions ...' option may be used for specifying the 3344 section to use. 3345 [Massimiliano Pala <madwolf@comune.modena.it>] 3346 3347 *) New OCSP utility. Allows OCSP requests to be generated or 3348 read. The request can be sent to a responder and the output 3349 parsed, outputed or printed in text form. Not complete yet: 3350 still needs to check the OCSP response validity. 3351 [Steve Henson] 3352 3353 *) New subcommands for 'openssl ca': 3354 'openssl ca -status <serial>' prints the status of the cert with 3355 the given serial number (according to the index file). 3356 'openssl ca -updatedb' updates the expiry status of certificates 3357 in the index file. 3358 [Massimiliano Pala <madwolf@comune.modena.it>] 3359 3360 *) New '-newreq-nodes' command option to CA.pl. This is like 3361 '-newreq', but calls 'openssl req' with the '-nodes' option 3362 so that the resulting key is not encrypted. 3363 [Damien Miller <djm@mindrot.org>] 3364 3365 *) New configuration for the GNU Hurd. 3366 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 3367 3368 *) Initial code to implement OCSP basic response verify. This 3369 is currently incomplete. Currently just finds the signer's 3370 certificate and verifies the signature on the response. 3371 [Steve Henson] 3372 3373 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 3374 value of OPENSSLDIR. This is available via the new '-d' option 3375 to 'openssl version', and is also included in 'openssl version -a'. 3376 [Bodo Moeller] 3377 3378 *) Allowing defining memory allocation callbacks that will be given 3379 file name and line number information in additional arguments 3380 (a const char* and an int). The basic functionality remains, as 3381 well as the original possibility to just replace malloc(), 3382 realloc() and free() by functions that do not know about these 3383 additional arguments. To register and find out the current 3384 settings for extended allocation functions, the following 3385 functions are provided: 3386 3387 CRYPTO_set_mem_ex_functions 3388 CRYPTO_set_locked_mem_ex_functions 3389 CRYPTO_get_mem_ex_functions 3390 CRYPTO_get_locked_mem_ex_functions 3391 3392 These work the same way as CRYPTO_set_mem_functions and friends. 3393 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 3394 extended allocation function is enabled. 3395 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 3396 a conventional allocation function is enabled. 3397 [Richard Levitte, Bodo Moeller] 3398 3399 *) Finish off removing the remaining LHASH function pointer casts. 3400 There should no longer be any prototype-casting required when using 3401 the LHASH abstraction, and any casts that remain are "bugs". See 3402 the callback types and macros at the head of lhash.h for details 3403 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 3404 [Geoff Thorpe] 3405 3406 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 3407 If /dev/[u]random devices are not available or do not return enough 3408 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 3409 be queried. 3410 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 3411 /etc/entropy will be queried once each in this sequence, quering stops 3412 when enough entropy was collected without querying more sockets. 3413 [Lutz Jaenicke] 3414 3415 *) Change the Unix RAND_poll() variant to be able to poll several 3416 random devices, as specified by DEVRANDOM, until a sufficient amount 3417 of data has been collected. We spend at most 10 ms on each file 3418 (select timeout) and read in non-blocking mode. DEVRANDOM now 3419 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 3420 (previously it was just the string "/dev/urandom"), so on typical 3421 platforms the 10 ms delay will never occur. 3422 Also separate out the Unix variant to its own file, rand_unix.c. 3423 For VMS, there's a currently-empty rand_vms.c. 3424 [Richard Levitte] 3425 3426 *) Move OCSP client related routines to ocsp_cl.c. These 3427 provide utility functions which an application needing 3428 to issue a request to an OCSP responder and analyse the 3429 response will typically need: as opposed to those which an 3430 OCSP responder itself would need which will be added later. 3431 3432 OCSP_request_sign() signs an OCSP request with an API similar 3433 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 3434 response. OCSP_response_get1_basic() extracts basic response 3435 from response. OCSP_resp_find_status(): finds and extracts status 3436 information from an OCSP_CERTID structure (which will be created 3437 when the request structure is built). These are built from lower 3438 level functions which work on OCSP_SINGLERESP structures but 3439 wont normally be used unless the application wishes to examine 3440 extensions in the OCSP response for example. 3441 3442 Replace nonce routines with a pair of functions. 3443 OCSP_request_add1_nonce() adds a nonce value and optionally 3444 generates a random value. OCSP_check_nonce() checks the 3445 validity of the nonce in an OCSP response. 3446 [Steve Henson] 3447 3448 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 3449 This doesn't copy the supplied OCSP_CERTID and avoids the 3450 need to free up the newly created id. Change return type 3451 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 3452 This can then be used to add extensions to the request. 3453 Deleted OCSP_request_new(), since most of its functionality 3454 is now in OCSP_REQUEST_new() (and the case insensitive name 3455 clash) apart from the ability to set the request name which 3456 will be added elsewhere. 3457 [Steve Henson] 3458 3459 *) Update OCSP API. Remove obsolete extensions argument from 3460 various functions. Extensions are now handled using the new 3461 OCSP extension code. New simple OCSP HTTP function which 3462 can be used to send requests and parse the response. 3463 [Steve Henson] 3464 3465 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 3466 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 3467 uses the special reorder version of SET OF to sort the attributes 3468 and reorder them to match the encoded order. This resolves a long 3469 standing problem: a verify on a PKCS7 structure just after signing 3470 it used to fail because the attribute order did not match the 3471 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 3472 it uses the received order. This is necessary to tolerate some broken 3473 software that does not order SET OF. This is handled by encoding 3474 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 3475 to produce the required SET OF. 3476 [Steve Henson] 3477 3478 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 3479 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 3480 files to get correct declarations of the ASN.1 item variables. 3481 [Richard Levitte] 3482 3483 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 3484 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 3485 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 3486 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 3487 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 3488 ASN1_ITEM and no wrapper functions. 3489 [Steve Henson] 3490 3491 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 3492 replace the old function pointer based I/O routines. Change most of 3493 the *_d2i_bio() and *_d2i_fp() functions to use these. 3494 [Steve Henson] 3495 3496 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 3497 lines, recognice more "algorithms" that can be deselected, and make 3498 it complain about algorithm deselection that isn't recognised. 3499 [Richard Levitte] 3500 3501 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 3502 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 3503 to use new functions. Add NO_ASN1_OLD which can be set to remove 3504 some old style ASN1 functions: this can be used to determine if old 3505 code will still work when these eventually go away. 3506 [Steve Henson] 3507 3508 *) New extension functions for OCSP structures, these follow the 3509 same conventions as certificates and CRLs. 3510 [Steve Henson] 3511 3512 *) New function X509V3_add1_i2d(). This automatically encodes and 3513 adds an extension. Its behaviour can be customised with various 3514 flags to append, replace or delete. Various wrappers added for 3515 certifcates and CRLs. 3516 [Steve Henson] 3517 3518 *) Fix to avoid calling the underlying ASN1 print routine when 3519 an extension cannot be parsed. Correct a typo in the 3520 OCSP_SERVICELOC extension. Tidy up print OCSP format. 3521 [Steve Henson] 3522 3523 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 3524 entries for variables. 3525 [Steve Henson] 3526 3527 *) Add functionality to apps/openssl.c for detecting locking 3528 problems: As the program is single-threaded, all we have 3529 to do is register a locking callback using an array for 3530 storing which locks are currently held by the program. 3531 [Bodo Moeller] 3532 3533 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 3534 SSL_get_ex_data_X509_STORE_idx(), which is used in 3535 ssl_verify_cert_chain() and thus can be called at any time 3536 during TLS/SSL handshakes so that thread-safety is essential. 3537 Unfortunately, the ex_data design is not at all suited 3538 for multi-threaded use, so it probably should be abolished. 3539 [Bodo Moeller] 3540 3541 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 3542 [Broadcom, tweaked and integrated by Geoff Thorpe] 3543 3544 *) Move common extension printing code to new function 3545 X509V3_print_extensions(). Reorganise OCSP print routines and 3546 implement some needed OCSP ASN1 functions. Add OCSP extensions. 3547 [Steve Henson] 3548 3549 *) New function X509_signature_print() to remove duplication in some 3550 print routines. 3551 [Steve Henson] 3552 3553 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 3554 set (this was treated exactly the same as SET OF previously). This 3555 is used to reorder the STACK representing the structure to match the 3556 encoding. This will be used to get round a problem where a PKCS7 3557 structure which was signed could not be verified because the STACK 3558 order did not reflect the encoded order. 3559 [Steve Henson] 3560 3561 *) Reimplement the OCSP ASN1 module using the new code. 3562 [Steve Henson] 3563 3564 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 3565 for its ASN1 operations. The old style function pointers still exist 3566 for now but they will eventually go away. 3567 [Steve Henson] 3568 3569 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 3570 completely replaces the old ASN1 functionality with a table driven 3571 encoder and decoder which interprets an ASN1_ITEM structure describing 3572 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 3573 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 3574 has also been converted to the new form. 3575 [Steve Henson] 3576 3577 *) Change BN_mod_exp_recp so that negative moduli are tolerated 3578 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 3579 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 3580 for negative moduli. 3581 [Bodo Moeller] 3582 3583 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 3584 of not touching the result's sign bit. 3585 [Bodo Moeller] 3586 3587 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 3588 set. 3589 [Bodo Moeller] 3590 3591 *) Changed the LHASH code to use prototypes for callbacks, and created 3592 macros to declare and implement thin (optionally static) functions 3593 that provide type-safety and avoid function pointer casting for the 3594 type-specific callbacks. 3595 [Geoff Thorpe] 3596 3597 *) Added Kerberos Cipher Suites to be used with TLS, as written in 3598 RFC 2712. 3599 [Veers Staats <staatsvr@asc.hpc.mil>, 3600 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 3601 3602 *) Reformat the FAQ so the different questions and answers can be divided 3603 in sections depending on the subject. 3604 [Richard Levitte] 3605 3606 *) Have the zlib compression code load ZLIB.DLL dynamically under 3607 Windows. 3608 [Richard Levitte] 3609 3610 *) New function BN_mod_sqrt for computing square roots modulo a prime 3611 (using the probabilistic Tonelli-Shanks algorithm unless 3612 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 3613 be handled deterministically). 3614 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3615 3616 *) Make BN_mod_inverse faster by explicitly handling small quotients 3617 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 3618 512 bits], about 30% for larger ones [1024 or 2048 bits].) 3619 [Bodo Moeller] 3620 3621 *) New function BN_kronecker. 3622 [Bodo Moeller] 3623 3624 *) Fix BN_gcd so that it works on negative inputs; the result is 3625 positive unless both parameters are zero. 3626 Previously something reasonably close to an infinite loop was 3627 possible because numbers could be growing instead of shrinking 3628 in the implementation of Euclid's algorithm. 3629 [Bodo Moeller] 3630 3631 *) Fix BN_is_word() and BN_is_one() macros to take into account the 3632 sign of the number in question. 3633 3634 Fix BN_is_word(a,w) to work correctly for w == 0. 3635 3636 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 3637 because its test if the absolute value of 'a' equals 'w'. 3638 Note that BN_abs_is_word does *not* handle w == 0 reliably; 3639 it exists mostly for use in the implementations of BN_is_zero(), 3640 BN_is_one(), and BN_is_word(). 3641 [Bodo Moeller] 3642 3643 *) New function BN_swap. 3644 [Bodo Moeller] 3645 3646 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 3647 the exponentiation functions are more likely to produce reasonable 3648 results on negative inputs. 3649 [Bodo Moeller] 3650 3651 *) Change BN_mod_mul so that the result is always non-negative. 3652 Previously, it could be negative if one of the factors was negative; 3653 I don't think anyone really wanted that behaviour. 3654 [Bodo Moeller] 3655 3656 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 3657 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 3658 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 3659 and add new functions: 3660 3661 BN_nnmod 3662 BN_mod_sqr 3663 BN_mod_add 3664 BN_mod_add_quick 3665 BN_mod_sub 3666 BN_mod_sub_quick 3667 BN_mod_lshift1 3668 BN_mod_lshift1_quick 3669 BN_mod_lshift 3670 BN_mod_lshift_quick 3671 3672 These functions always generate non-negative results. 3673 3674 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 3675 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 3676 3677 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 3678 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 3679 be reduced modulo m. 3680 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3681 3682#if 0 3683 The following entry accidentily appeared in the CHANGES file 3684 distributed with OpenSSL 0.9.7. The modifications described in 3685 it do *not* apply to OpenSSL 0.9.7. 3686 3687 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3688 was actually never needed) and in BN_mul(). The removal in BN_mul() 3689 required a small change in bn_mul_part_recursive() and the addition 3690 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3691 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3692 bn_sub_words() and bn_add_words() except they take arrays with 3693 differing sizes. 3694 [Richard Levitte] 3695#endif 3696 3697 *) In 'openssl passwd', verify passwords read from the terminal 3698 unless the '-salt' option is used (which usually means that 3699 verification would just waste user's time since the resulting 3700 hash is going to be compared with some given password hash) 3701 or the new '-noverify' option is used. 3702 3703 This is an incompatible change, but it does not affect 3704 non-interactive use of 'openssl passwd' (passwords on the command 3705 line, '-stdin' option, '-in ...' option) and thus should not 3706 cause any problems. 3707 [Bodo Moeller] 3708 3709 *) Remove all references to RSAref, since there's no more need for it. 3710 [Richard Levitte] 3711 3712 *) Make DSO load along a path given through an environment variable 3713 (SHLIB_PATH) with shl_load(). 3714 [Richard Levitte] 3715 3716 *) Constify the ENGINE code as a result of BIGNUM constification. 3717 Also constify the RSA code and most things related to it. In a 3718 few places, most notable in the depth of the ASN.1 code, ugly 3719 casts back to non-const were required (to be solved at a later 3720 time) 3721 [Richard Levitte] 3722 3723 *) Make it so the openssl application has all engines loaded by default. 3724 [Richard Levitte] 3725 3726 *) Constify the BIGNUM routines a little more. 3727 [Richard Levitte] 3728 3729 *) Add the following functions: 3730 3731 ENGINE_load_cswift() 3732 ENGINE_load_chil() 3733 ENGINE_load_atalla() 3734 ENGINE_load_nuron() 3735 ENGINE_load_builtin_engines() 3736 3737 That way, an application can itself choose if external engines that 3738 are built-in in OpenSSL shall ever be used or not. The benefit is 3739 that applications won't have to be linked with libdl or other dso 3740 libraries unless it's really needed. 3741 3742 Changed 'openssl engine' to load all engines on demand. 3743 Changed the engine header files to avoid the duplication of some 3744 declarations (they differed!). 3745 [Richard Levitte] 3746 3747 *) 'openssl engine' can now list capabilities. 3748 [Richard Levitte] 3749 3750 *) Better error reporting in 'openssl engine'. 3751 [Richard Levitte] 3752 3753 *) Never call load_dh_param(NULL) in s_server. 3754 [Bodo Moeller] 3755 3756 *) Add engine application. It can currently list engines by name and 3757 identity, and test if they are actually available. 3758 [Richard Levitte] 3759 3760 *) Improve RPM specification file by forcing symbolic linking and making 3761 sure the installed documentation is also owned by root.root. 3762 [Damien Miller <djm@mindrot.org>] 3763 3764 *) Give the OpenSSL applications more possibilities to make use of 3765 keys (public as well as private) handled by engines. 3766 [Richard Levitte] 3767 3768 *) Add OCSP code that comes from CertCo. 3769 [Richard Levitte] 3770 3771 *) Add VMS support for the Rijndael code. 3772 [Richard Levitte] 3773 3774 *) Added untested support for Nuron crypto accelerator. 3775 [Ben Laurie] 3776 3777 *) Add support for external cryptographic devices. This code was 3778 previously distributed separately as the "engine" branch. 3779 [Geoff Thorpe, Richard Levitte] 3780 3781 *) Rework the filename-translation in the DSO code. It is now possible to 3782 have far greater control over how a "name" is turned into a filename 3783 depending on the operating environment and any oddities about the 3784 different shared library filenames on each system. 3785 [Geoff Thorpe] 3786 3787 *) Support threads on FreeBSD-elf in Configure. 3788 [Richard Levitte] 3789 3790 *) Fix for SHA1 assembly problem with MASM: it produces 3791 warnings about corrupt line number information when assembling 3792 with debugging information. This is caused by the overlapping 3793 of two sections. 3794 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 3795 3796 *) NCONF changes. 3797 NCONF_get_number() has no error checking at all. As a replacement, 3798 NCONF_get_number_e() is defined (_e for "error checking") and is 3799 promoted strongly. The old NCONF_get_number is kept around for 3800 binary backward compatibility. 3801 Make it possible for methods to load from something other than a BIO, 3802 by providing a function pointer that is given a name instead of a BIO. 3803 For example, this could be used to load configuration data from an 3804 LDAP server. 3805 [Richard Levitte] 3806 3807 *) Fix for non blocking accept BIOs. Added new I/O special reason 3808 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 3809 with non blocking I/O was not possible because no retry code was 3810 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 3811 this case. 3812 [Steve Henson] 3813 3814 *) Added the beginnings of Rijndael support. 3815 [Ben Laurie] 3816 3817 *) Fix for bug in DirectoryString mask setting. Add support for 3818 X509_NAME_print_ex() in 'req' and X509_print_ex() function 3819 to allow certificate printing to more controllable, additional 3820 'certopt' option to 'x509' to allow new printing options to be 3821 set. 3822 [Steve Henson] 3823 3824 *) Clean old EAY MD5 hack from e_os.h. 3825 [Richard Levitte] 3826 3827 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 3828 3829 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 3830 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 3831 [Joe Orton, Steve Henson] 3832 3833 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 3834 3835 *) Fix additional bug revealed by the NISCC test suite: 3836 3837 Stop bug triggering large recursion when presented with 3838 certain ASN.1 tags (CVE-2003-0851) 3839 [Steve Henson] 3840 3841 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 3842 3843 *) Fix various bugs revealed by running the NISCC test suite: 3844 3845 Stop out of bounds reads in the ASN1 code when presented with 3846 invalid tags (CVE-2003-0543 and CVE-2003-0544). 3847 3848 If verify callback ignores invalid public key errors don't try to check 3849 certificate signature with the NULL public key. 3850 3851 [Steve Henson] 3852 3853 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 3854 if the server requested one: as stated in TLS 1.0 and SSL 3.0 3855 specifications. 3856 [Steve Henson] 3857 3858 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 3859 extra data after the compression methods not only for TLS 1.0 3860 but also for SSL 3.0 (as required by the specification). 3861 [Bodo Moeller; problem pointed out by Matthias Loepfe] 3862 3863 *) Change X509_certificate_type() to mark the key as exported/exportable 3864 when it's 512 *bits* long, not 512 bytes. 3865 [Richard Levitte] 3866 3867 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 3868 3869 *) Countermeasure against the Klima-Pokorny-Rosa extension of 3870 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 3871 a protocol version number mismatch like a decryption error 3872 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 3873 [Bodo Moeller] 3874 3875 *) Turn on RSA blinding by default in the default implementation 3876 to avoid a timing attack. Applications that don't want it can call 3877 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 3878 They would be ill-advised to do so in most cases. 3879 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 3880 3881 *) Change RSA blinding code so that it works when the PRNG is not 3882 seeded (in this case, the secret RSA exponent is abused as 3883 an unpredictable seed -- if it is not unpredictable, there 3884 is no point in blinding anyway). Make RSA blinding thread-safe 3885 by remembering the creator's thread ID in rsa->blinding and 3886 having all other threads use local one-time blinding factors 3887 (this requires more computation than sharing rsa->blinding, but 3888 avoids excessive locking; and if an RSA object is not shared 3889 between threads, blinding will still be very fast). 3890 [Bodo Moeller] 3891 3892 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 3893 3894 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 3895 via timing by performing a MAC computation even if incorrrect 3896 block cipher padding has been found. This is a countermeasure 3897 against active attacks where the attacker has to distinguish 3898 between bad padding and a MAC verification error. (CVE-2003-0078) 3899 3900 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 3901 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 3902 Martin Vuagnoux (EPFL, Ilion)] 3903 3904 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 3905 3906 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 3907 memory from it's contents. This is done with a counter that will 3908 place alternating values in each byte. This can be used to solve 3909 two issues: 1) the removal of calls to memset() by highly optimizing 3910 compilers, and 2) cleansing with other values than 0, since those can 3911 be read through on certain media, for example a swap space on disk. 3912 [Geoff Thorpe] 3913 3914 *) Bugfix: client side session caching did not work with external caching, 3915 because the session->cipher setting was not restored when reloading 3916 from the external cache. This problem was masked, when 3917 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 3918 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 3919 [Lutz Jaenicke] 3920 3921 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 3922 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 3923 [Zeev Lieber <zeev-l@yahoo.com>] 3924 3925 *) Undo an undocumented change introduced in 0.9.6e which caused 3926 repeated calls to OpenSSL_add_all_ciphers() and 3927 OpenSSL_add_all_digests() to be ignored, even after calling 3928 EVP_cleanup(). 3929 [Richard Levitte] 3930 3931 *) Change the default configuration reader to deal with last line not 3932 being properly terminated. 3933 [Richard Levitte] 3934 3935 *) Change X509_NAME_cmp() so it applies the special rules on handling 3936 DN values that are of type PrintableString, as well as RDNs of type 3937 emailAddress where the value has the type ia5String. 3938 [stefank@valicert.com via Richard Levitte] 3939 3940 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 3941 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 3942 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 3943 the bitwise-OR of the two for use by the majority of applications 3944 wanting this behaviour, and update the docs. The documented 3945 behaviour and actual behaviour were inconsistent and had been 3946 changing anyway, so this is more a bug-fix than a behavioural 3947 change. 3948 [Geoff Thorpe, diagnosed by Nadav Har'El] 3949 3950 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 3951 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 3952 [Bodo Moeller] 3953 3954 *) Fix initialization code race conditions in 3955 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 3956 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 3957 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 3958 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 3959 ssl2_get_cipher_by_char(), 3960 ssl3_get_cipher_by_char(). 3961 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 3962 3963 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 3964 the cached sessions are flushed, as the remove_cb() might use ex_data 3965 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 3966 (see [openssl.org #212]). 3967 [Geoff Thorpe, Lutz Jaenicke] 3968 3969 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 3970 length, instead of the encoding length to d2i_ASN1_OBJECT. 3971 [Steve Henson] 3972 3973 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 3974 3975 *) [In 0.9.6g-engine release:] 3976 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 3977 [Lynn Gazis <lgazis@rainbow.com>] 3978 3979 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 3980 3981 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 3982 and get fix the header length calculation. 3983 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 3984 Alon Kantor <alonk@checkpoint.com> (and others), 3985 Steve Henson] 3986 3987 *) Use proper error handling instead of 'assertions' in buffer 3988 overflow checks added in 0.9.6e. This prevents DoS (the 3989 assertions could call abort()). 3990 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 3991 3992 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 3993 3994 *) Add various sanity checks to asn1_get_length() to reject 3995 the ASN1 length bytes if they exceed sizeof(long), will appear 3996 negative or the content length exceeds the length of the 3997 supplied buffer. 3998 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 3999 4000 *) Fix cipher selection routines: ciphers without encryption had no flags 4001 for the cipher strength set and where therefore not handled correctly 4002 by the selection routines (PR #130). 4003 [Lutz Jaenicke] 4004 4005 *) Fix EVP_dsa_sha macro. 4006 [Nils Larsch] 4007 4008 *) New option 4009 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 4010 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 4011 that was added in OpenSSL 0.9.6d. 4012 4013 As the countermeasure turned out to be incompatible with some 4014 broken SSL implementations, the new option is part of SSL_OP_ALL. 4015 SSL_OP_ALL is usually employed when compatibility with weird SSL 4016 implementations is desired (e.g. '-bugs' option to 's_client' and 4017 's_server'), so the new option is automatically set in many 4018 applications. 4019 [Bodo Moeller] 4020 4021 *) Changes in security patch: 4022 4023 Changes marked "(CHATS)" were sponsored by the Defense Advanced 4024 Research Projects Agency (DARPA) and Air Force Research Laboratory, 4025 Air Force Materiel Command, USAF, under agreement number 4026 F30602-01-2-0537. 4027 4028 *) Add various sanity checks to asn1_get_length() to reject 4029 the ASN1 length bytes if they exceed sizeof(long), will appear 4030 negative or the content length exceeds the length of the 4031 supplied buffer. (CVE-2002-0659) 4032 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4033 4034 *) Assertions for various potential buffer overflows, not known to 4035 happen in practice. 4036 [Ben Laurie (CHATS)] 4037 4038 *) Various temporary buffers to hold ASCII versions of integers were 4039 too small for 64 bit platforms. (CVE-2002-0655) 4040 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 4041 4042 *) Remote buffer overflow in SSL3 protocol - an attacker could 4043 supply an oversized session ID to a client. (CVE-2002-0656) 4044 [Ben Laurie (CHATS)] 4045 4046 *) Remote buffer overflow in SSL2 protocol - an attacker could 4047 supply an oversized client master key. (CVE-2002-0656) 4048 [Ben Laurie (CHATS)] 4049 4050 Changes between 0.9.6c and 0.9.6d [9 May 2002] 4051 4052 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 4053 encoded as NULL) with id-dsa-with-sha1. 4054 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 4055 4056 *) Check various X509_...() return values in apps/req.c. 4057 [Nils Larsch <nla@trustcenter.de>] 4058 4059 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 4060 an end-of-file condition would erronously be flagged, when the CRLF 4061 was just at the end of a processed block. The bug was discovered when 4062 processing data through a buffering memory BIO handing the data to a 4063 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 4064 <ptsekov@syntrex.com> and Nedelcho Stanev. 4065 [Lutz Jaenicke] 4066 4067 *) Implement a countermeasure against a vulnerability recently found 4068 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 4069 before application data chunks to avoid the use of known IVs 4070 with data potentially chosen by the attacker. 4071 [Bodo Moeller] 4072 4073 *) Fix length checks in ssl3_get_client_hello(). 4074 [Bodo Moeller] 4075 4076 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 4077 to prevent ssl3_read_internal() from incorrectly assuming that 4078 ssl3_read_bytes() found application data while handshake 4079 processing was enabled when in fact s->s3->in_read_app_data was 4080 merely automatically cleared during the initial handshake. 4081 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 4082 4083 *) Fix object definitions for Private and Enterprise: they were not 4084 recognized in their shortname (=lowercase) representation. Extend 4085 obj_dat.pl to issue an error when using undefined keywords instead 4086 of silently ignoring the problem (Svenning Sorensen 4087 <sss@sss.dnsalias.net>). 4088 [Lutz Jaenicke] 4089 4090 *) Fix DH_generate_parameters() so that it works for 'non-standard' 4091 generators, i.e. generators other than 2 and 5. (Previously, the 4092 code did not properly initialise the 'add' and 'rem' values to 4093 BN_generate_prime().) 4094 4095 In the new general case, we do not insist that 'generator' is 4096 actually a primitive root: This requirement is rather pointless; 4097 a generator of the order-q subgroup is just as good, if not 4098 better. 4099 [Bodo Moeller] 4100 4101 *) Map new X509 verification errors to alerts. Discovered and submitted by 4102 Tom Wu <tom@arcot.com>. 4103 [Lutz Jaenicke] 4104 4105 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 4106 returning non-zero before the data has been completely received 4107 when using non-blocking I/O. 4108 [Bodo Moeller; problem pointed out by John Hughes] 4109 4110 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 4111 [Ben Laurie, Lutz Jaenicke] 4112 4113 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 4114 Yoram Zahavi <YoramZ@gilian.com>). 4115 [Lutz Jaenicke] 4116 4117 *) Add information about CygWin 1.3 and on, and preserve proper 4118 configuration for the versions before that. 4119 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4120 4121 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 4122 check whether we deal with a copy of a session and do not delete from 4123 the cache in this case. Problem reported by "Izhar Shoshani Levi" 4124 <izhar@checkpoint.com>. 4125 [Lutz Jaenicke] 4126 4127 *) Do not store session data into the internal session cache, if it 4128 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 4129 flag is set). Proposed by Aslam <aslam@funk.com>. 4130 [Lutz Jaenicke] 4131 4132 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 4133 value is 0. 4134 [Richard Levitte] 4135 4136 *) [In 0.9.6d-engine release:] 4137 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4138 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4139 4140 *) Add the configuration target linux-s390x. 4141 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 4142 4143 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 4144 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 4145 variable as an indication that a ClientHello message has been 4146 received. As the flag value will be lost between multiple 4147 invocations of ssl3_accept when using non-blocking I/O, the 4148 function may not be aware that a handshake has actually taken 4149 place, thus preventing a new session from being added to the 4150 session cache. 4151 4152 To avoid this problem, we now set s->new_session to 2 instead of 4153 using a local variable. 4154 [Lutz Jaenicke, Bodo Moeller] 4155 4156 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 4157 if the SSL_R_LENGTH_MISMATCH error is detected. 4158 [Geoff Thorpe, Bodo Moeller] 4159 4160 *) New 'shared_ldflag' column in Configure platform table. 4161 [Richard Levitte] 4162 4163 *) Fix EVP_CIPHER_mode macro. 4164 ["Dan S. Camper" <dan@bti.net>] 4165 4166 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 4167 type, we must throw them away by setting rr->length to 0. 4168 [D P Chang <dpc@qualys.com>] 4169 4170 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 4171 4172 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 4173 <Dominikus.Scherkl@biodata.com>. (The previous implementation 4174 worked incorrectly for those cases where range = 10..._2 and 4175 3*range is two bits longer than range.) 4176 [Bodo Moeller] 4177 4178 *) Only add signing time to PKCS7 structures if it is not already 4179 present. 4180 [Steve Henson] 4181 4182 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 4183 OBJ_ld_ce should be OBJ_id_ce. 4184 Also some ip-pda OIDs in crypto/objects/objects.txt were 4185 incorrect (cf. RFC 3039). 4186 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 4187 4188 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 4189 returns early because it has nothing to do. 4190 [Andy Schneider <andy.schneider@bjss.co.uk>] 4191 4192 *) [In 0.9.6c-engine release:] 4193 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 4194 [Andy Schneider <andy.schneider@bjss.co.uk>] 4195 4196 *) [In 0.9.6c-engine release:] 4197 Add support for Cryptographic Appliance's keyserver technology. 4198 (Use engine 'keyclient') 4199 [Cryptographic Appliances and Geoff Thorpe] 4200 4201 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 4202 is called via tools/c89.sh because arguments have to be 4203 rearranged (all '-L' options must appear before the first object 4204 modules). 4205 [Richard Shapiro <rshapiro@abinitio.com>] 4206 4207 *) [In 0.9.6c-engine release:] 4208 Add support for Broadcom crypto accelerator cards, backported 4209 from 0.9.7. 4210 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 4211 4212 *) [In 0.9.6c-engine release:] 4213 Add support for SureWare crypto accelerator cards from 4214 Baltimore Technologies. (Use engine 'sureware') 4215 [Baltimore Technologies and Mark Cox] 4216 4217 *) [In 0.9.6c-engine release:] 4218 Add support for crypto accelerator cards from Accelerated 4219 Encryption Processing, www.aep.ie. (Use engine 'aep') 4220 [AEP Inc. and Mark Cox] 4221 4222 *) Add a configuration entry for gcc on UnixWare. 4223 [Gary Benson <gbenson@redhat.com>] 4224 4225 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 4226 messages are stored in a single piece (fixed-length part and 4227 variable-length part combined) and fix various bugs found on the way. 4228 [Bodo Moeller] 4229 4230 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 4231 instead. BIO_gethostbyname() does not know what timeouts are 4232 appropriate, so entries would stay in cache even when they have 4233 become invalid. 4234 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 4235 4236 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 4237 faced with a pathologically small ClientHello fragment that does 4238 not contain client_version: Instead of aborting with an error, 4239 simply choose the highest available protocol version (i.e., 4240 TLS 1.0 unless it is disabled). In practice, ClientHello 4241 messages are never sent like this, but this change gives us 4242 strictly correct behaviour at least for TLS. 4243 [Bodo Moeller] 4244 4245 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 4246 never resets s->method to s->ctx->method when called from within 4247 one of the SSL handshake functions. 4248 [Bodo Moeller; problem pointed out by Niko Baric] 4249 4250 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 4251 (sent using the client's version number) if client_version is 4252 smaller than the protocol version in use. Also change 4253 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 4254 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 4255 the client will at least see that alert. 4256 [Bodo Moeller] 4257 4258 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 4259 correctly. 4260 [Bodo Moeller] 4261 4262 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 4263 client receives HelloRequest while in a handshake. 4264 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 4265 4266 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 4267 should end in 'break', not 'goto end' which circuments various 4268 cleanups done in state SSL_ST_OK. But session related stuff 4269 must be disabled for SSL_ST_OK in the case that we just sent a 4270 HelloRequest. 4271 4272 Also avoid some overhead by not calling ssl_init_wbio_buffer() 4273 before just sending a HelloRequest. 4274 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 4275 4276 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 4277 reveal whether illegal block cipher padding was found or a MAC 4278 verification error occured. (Neither SSLerr() codes nor alerts 4279 are directly visible to potential attackers, but the information 4280 may leak via logfiles.) 4281 4282 Similar changes are not required for the SSL 2.0 implementation 4283 because the number of padding bytes is sent in clear for SSL 2.0, 4284 and the extra bytes are just ignored. However ssl/s2_pkt.c 4285 failed to verify that the purported number of padding bytes is in 4286 the legal range. 4287 [Bodo Moeller] 4288 4289 *) Add OpenUNIX-8 support including shared libraries 4290 (Boyd Lynn Gerber <gerberb@zenez.com>). 4291 [Lutz Jaenicke] 4292 4293 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 4294 'wristwatch attack' using huge encoding parameters (cf. 4295 James H. Manger's CRYPTO 2001 paper). Note that the 4296 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 4297 encoding parameters and hence was not vulnerable. 4298 [Bodo Moeller] 4299 4300 *) BN_sqr() bug fix. 4301 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 4302 4303 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 4304 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 4305 followed by modular reduction. 4306 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 4307 4308 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 4309 equivalent based on BN_pseudo_rand() instead of BN_rand(). 4310 [Bodo Moeller] 4311 4312 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 4313 This function was broken, as the check for a new client hello message 4314 to handle SGC did not allow these large messages. 4315 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 4316 [Lutz Jaenicke] 4317 4318 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 4319 [Lutz Jaenicke] 4320 4321 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 4322 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 4323 [Lutz Jaenicke] 4324 4325 *) Rework the configuration and shared library support for Tru64 Unix. 4326 The configuration part makes use of modern compiler features and 4327 still retains old compiler behavior for those that run older versions 4328 of the OS. The shared library support part includes a variant that 4329 uses the RPATH feature, and is available through the special 4330 configuration target "alpha-cc-rpath", which will never be selected 4331 automatically. 4332 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 4333 4334 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 4335 with the same message size as in ssl3_get_certificate_request(). 4336 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 4337 messages might inadvertently be reject as too long. 4338 [Petr Lampa <lampa@fee.vutbr.cz>] 4339 4340 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 4341 [Andy Polyakov] 4342 4343 *) Modified SSL library such that the verify_callback that has been set 4344 specificly for an SSL object with SSL_set_verify() is actually being 4345 used. Before the change, a verify_callback set with this function was 4346 ignored and the verify_callback() set in the SSL_CTX at the time of 4347 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 4348 to allow the necessary settings. 4349 [Lutz Jaenicke] 4350 4351 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 4352 explicitly to NULL, as at least on Solaris 8 this seems not always to be 4353 done automatically (in contradiction to the requirements of the C 4354 standard). This made problems when used from OpenSSH. 4355 [Lutz Jaenicke] 4356 4357 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 4358 dh->length and always used 4359 4360 BN_rand_range(priv_key, dh->p). 4361 4362 BN_rand_range() is not necessary for Diffie-Hellman, and this 4363 specific range makes Diffie-Hellman unnecessarily inefficient if 4364 dh->length (recommended exponent length) is much smaller than the 4365 length of dh->p. We could use BN_rand_range() if the order of 4366 the subgroup was stored in the DH structure, but we only have 4367 dh->length. 4368 4369 So switch back to 4370 4371 BN_rand(priv_key, l, ...) 4372 4373 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 4374 otherwise. 4375 [Bodo Moeller] 4376 4377 *) In 4378 4379 RSA_eay_public_encrypt 4380 RSA_eay_private_decrypt 4381 RSA_eay_private_encrypt (signing) 4382 RSA_eay_public_decrypt (signature verification) 4383 4384 (default implementations for RSA_public_encrypt, 4385 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 4386 always reject numbers >= n. 4387 [Bodo Moeller] 4388 4389 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 4390 to synchronize access to 'locking_thread'. This is necessary on 4391 systems where access to 'locking_thread' (an 'unsigned long' 4392 variable) is not atomic. 4393 [Bodo Moeller] 4394 4395 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 4396 *before* setting the 'crypto_lock_rand' flag. The previous code had 4397 a race condition if 0 is a valid thread ID. 4398 [Travis Vitek <vitek@roguewave.com>] 4399 4400 *) Add support for shared libraries under Irix. 4401 [Albert Chin-A-Young <china@thewrittenword.com>] 4402 4403 *) Add configuration option to build on Linux on both big-endian and 4404 little-endian MIPS. 4405 [Ralf Baechle <ralf@uni-koblenz.de>] 4406 4407 *) Add the possibility to create shared libraries on HP-UX. 4408 [Richard Levitte] 4409 4410 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 4411 4412 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 4413 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 4414 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 4415 PRNG state recovery was possible based on the output of 4416 one PRNG request appropriately sized to gain knowledge on 4417 'md' followed by enough consecutive 1-byte PRNG requests 4418 to traverse all of 'state'. 4419 4420 1. When updating 'md_local' (the current thread's copy of 'md') 4421 during PRNG output generation, hash all of the previous 4422 'md_local' value, not just the half used for PRNG output. 4423 4424 2. Make the number of bytes from 'state' included into the hash 4425 independent from the number of PRNG bytes requested. 4426 4427 The first measure alone would be sufficient to avoid 4428 Markku-Juhani's attack. (Actually it had never occurred 4429 to me that the half of 'md_local' used for chaining was the 4430 half from which PRNG output bytes were taken -- I had always 4431 assumed that the secret half would be used.) The second 4432 measure makes sure that additional data from 'state' is never 4433 mixed into 'md_local' in small portions; this heuristically 4434 further strengthens the PRNG. 4435 [Bodo Moeller] 4436 4437 *) Fix crypto/bn/asm/mips3.s. 4438 [Andy Polyakov] 4439 4440 *) When only the key is given to "enc", the IV is undefined. Print out 4441 an error message in this case. 4442 [Lutz Jaenicke] 4443 4444 *) Handle special case when X509_NAME is empty in X509 printing routines. 4445 [Steve Henson] 4446 4447 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 4448 positive and less than q. 4449 [Bodo Moeller] 4450 4451 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 4452 used: it isn't thread safe and the add_lock_callback should handle 4453 that itself. 4454 [Paul Rose <Paul.Rose@bridge.com>] 4455 4456 *) Verify that incoming data obeys the block size in 4457 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 4458 [Bodo Moeller] 4459 4460 *) Fix OAEP check. 4461 [Ulf M�ller, Bodo M�ller] 4462 4463 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 4464 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 4465 when fixing the server behaviour for backwards-compatible 'client 4466 hello' messages. (Note that the attack is impractical against 4467 SSL 3.0 and TLS 1.0 anyway because length and version checking 4468 means that the probability of guessing a valid ciphertext is 4469 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 4470 paper.) 4471 4472 Before 0.9.5, the countermeasure (hide the error by generating a 4473 random 'decryption result') did not work properly because 4474 ERR_clear_error() was missing, meaning that SSL_get_error() would 4475 detect the supposedly ignored error. 4476 4477 Both problems are now fixed. 4478 [Bodo Moeller] 4479 4480 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 4481 (previously it was 1024). 4482 [Bodo Moeller] 4483 4484 *) Fix for compatibility mode trust settings: ignore trust settings 4485 unless some valid trust or reject settings are present. 4486 [Steve Henson] 4487 4488 *) Fix for blowfish EVP: its a variable length cipher. 4489 [Steve Henson] 4490 4491 *) Fix various bugs related to DSA S/MIME verification. Handle missing 4492 parameters in DSA public key structures and return an error in the 4493 DSA routines if parameters are absent. 4494 [Steve Henson] 4495 4496 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 4497 in the current directory if neither $RANDFILE nor $HOME was set. 4498 RAND_file_name() in 0.9.6a returned NULL in this case. This has 4499 caused some confusion to Windows users who haven't defined $HOME. 4500 Thus RAND_file_name() is changed again: e_os.h can define a 4501 DEFAULT_HOME, which will be used if $HOME is not set. 4502 For Windows, we use "C:"; on other platforms, we still require 4503 environment variables. 4504 4505 *) Move 'if (!initialized) RAND_poll()' into regions protected by 4506 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 4507 having multiple threads call RAND_poll() concurrently. 4508 [Bodo Moeller] 4509 4510 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 4511 combination of a flag and a thread ID variable. 4512 Otherwise while one thread is in ssleay_rand_bytes (which sets the 4513 flag), *other* threads can enter ssleay_add_bytes without obeying 4514 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 4515 that they do not hold after the first thread unsets add_do_not_lock). 4516 [Bodo Moeller] 4517 4518 *) Change bctest again: '-x' expressions are not available in all 4519 versions of 'test'. 4520 [Bodo Moeller] 4521 4522 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 4523 4524 *) Fix a couple of memory leaks in PKCS7_dataDecode() 4525 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 4526 4527 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 4528 the default extension for executables, if any. Also, make the perl 4529 scripts that use symlink() to test if it really exists and use "cp" 4530 if it doesn't. All this made OpenSSL compilable and installable in 4531 CygWin. 4532 [Richard Levitte] 4533 4534 *) Fix for asn1_GetSequence() for indefinite length constructed data. 4535 If SEQUENCE is length is indefinite just set c->slen to the total 4536 amount of data available. 4537 [Steve Henson, reported by shige@FreeBSD.org] 4538 [This change does not apply to 0.9.7.] 4539 4540 *) Change bctest to avoid here-documents inside command substitution 4541 (workaround for FreeBSD /bin/sh bug). 4542 For compatibility with Ultrix, avoid shell functions (introduced 4543 in the bctest version that searches along $PATH). 4544 [Bodo Moeller] 4545 4546 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 4547 with des_encrypt() defined on some operating systems, like Solaris 4548 and UnixWare. 4549 [Richard Levitte] 4550 4551 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 4552 On the Importance of Eliminating Errors in Cryptographic 4553 Computations, J. Cryptology 14 (2001) 2, 101-119, 4554 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 4555 [Ulf Moeller] 4556 4557 *) MIPS assembler BIGNUM division bug fix. 4558 [Andy Polyakov] 4559 4560 *) Disabled incorrect Alpha assembler code. 4561 [Richard Levitte] 4562 4563 *) Fix PKCS#7 decode routines so they correctly update the length 4564 after reading an EOC for the EXPLICIT tag. 4565 [Steve Henson] 4566 [This change does not apply to 0.9.7.] 4567 4568 *) Fix bug in PKCS#12 key generation routines. This was triggered 4569 if a 3DES key was generated with a 0 initial byte. Include 4570 PKCS12_BROKEN_KEYGEN compilation option to retain the old 4571 (but broken) behaviour. 4572 [Steve Henson] 4573 4574 *) Enhance bctest to search for a working bc along $PATH and print 4575 it when found. 4576 [Tim Rice <tim@multitalents.net> via Richard Levitte] 4577 4578 *) Fix memory leaks in err.c: free err_data string if necessary; 4579 don't write to the wrong index in ERR_set_error_data. 4580 [Bodo Moeller] 4581 4582 *) Implement ssl23_peek (analogous to ssl23_read), which previously 4583 did not exist. 4584 [Bodo Moeller] 4585 4586 *) Replace rdtsc with _emit statements for VC++ version 5. 4587 [Jeremy Cooper <jeremy@baymoo.org>] 4588 4589 *) Make it possible to reuse SSLv2 sessions. 4590 [Richard Levitte] 4591 4592 *) In copy_email() check for >= 0 as a return value for 4593 X509_NAME_get_index_by_NID() since 0 is a valid index. 4594 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 4595 4596 *) Avoid coredump with unsupported or invalid public keys by checking if 4597 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 4598 PKCS7_verify() fails with non detached data. 4599 [Steve Henson] 4600 4601 *) Don't use getenv in library functions when run as setuid/setgid. 4602 New function OPENSSL_issetugid(). 4603 [Ulf Moeller] 4604 4605 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 4606 due to incorrect handling of multi-threading: 4607 4608 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 4609 4610 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 4611 4612 3. Count how many times MemCheck_off() has been called so that 4613 nested use can be treated correctly. This also avoids 4614 inband-signalling in the previous code (which relied on the 4615 assumption that thread ID 0 is impossible). 4616 [Bodo Moeller] 4617 4618 *) Add "-rand" option also to s_client and s_server. 4619 [Lutz Jaenicke] 4620 4621 *) Fix CPU detection on Irix 6.x. 4622 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 4623 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4624 4625 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 4626 was empty. 4627 [Steve Henson] 4628 [This change does not apply to 0.9.7.] 4629 4630 *) Use the cached encoding of an X509_NAME structure rather than 4631 copying it. This is apparently the reason for the libsafe "errors" 4632 but the code is actually correct. 4633 [Steve Henson] 4634 4635 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 4636 Bleichenbacher's DSA attack. 4637 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 4638 to be set and top=0 forces the highest bit to be set; top=-1 is new 4639 and leaves the highest bit random. 4640 [Ulf Moeller, Bodo Moeller] 4641 4642 *) In the NCONF_...-based implementations for CONF_... queries 4643 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 4644 a temporary CONF structure with the data component set to NULL 4645 (which gives segmentation faults in lh_retrieve). 4646 Instead, use NULL for the CONF pointer in CONF_get_string and 4647 CONF_get_number (which may use environment variables) and directly 4648 return NULL from CONF_get_section. 4649 [Bodo Moeller] 4650 4651 *) Fix potential buffer overrun for EBCDIC. 4652 [Ulf Moeller] 4653 4654 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 4655 keyUsage if basicConstraints absent for a CA. 4656 [Steve Henson] 4657 4658 *) Make SMIME_write_PKCS7() write mail header values with a format that 4659 is more generally accepted (no spaces before the semicolon), since 4660 some programs can't parse those values properly otherwise. Also make 4661 sure BIO's that break lines after each write do not create invalid 4662 headers. 4663 [Richard Levitte] 4664 4665 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 4666 macros previously used would not encode an empty SEQUENCE OF 4667 and break the signature. 4668 [Steve Henson] 4669 [This change does not apply to 0.9.7.] 4670 4671 *) Zero the premaster secret after deriving the master secret in 4672 DH ciphersuites. 4673 [Steve Henson] 4674 4675 *) Add some EVP_add_digest_alias registrations (as found in 4676 OpenSSL_add_all_digests()) to SSL_library_init() 4677 aka OpenSSL_add_ssl_algorithms(). This provides improved 4678 compatibility with peers using X.509 certificates 4679 with unconventional AlgorithmIdentifier OIDs. 4680 [Bodo Moeller] 4681 4682 *) Fix for Irix with NO_ASM. 4683 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4684 4685 *) ./config script fixes. 4686 [Ulf Moeller, Richard Levitte] 4687 4688 *) Fix 'openssl passwd -1'. 4689 [Bodo Moeller] 4690 4691 *) Change PKCS12_key_gen_asc() so it can cope with non null 4692 terminated strings whose length is passed in the passlen 4693 parameter, for example from PEM callbacks. This was done 4694 by adding an extra length parameter to asc2uni(). 4695 [Steve Henson, reported by <oddissey@samsung.co.kr>] 4696 4697 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 4698 call failed, free the DSA structure. 4699 [Bodo Moeller] 4700 4701 *) Fix to uni2asc() to cope with zero length Unicode strings. 4702 These are present in some PKCS#12 files. 4703 [Steve Henson] 4704 4705 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 4706 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 4707 when writing a 32767 byte record. 4708 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 4709 4710 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 4711 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 4712 4713 (RSA objects have a reference count access to which is protected 4714 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 4715 so they are meant to be shared between threads.) 4716 [Bodo Moeller, Geoff Thorpe; original patch submitted by 4717 "Reddie, Steven" <Steven.Reddie@ca.com>] 4718 4719 *) Fix a deadlock in CRYPTO_mem_leaks(). 4720 [Bodo Moeller] 4721 4722 *) Use better test patterns in bntest. 4723 [Ulf M�ller] 4724 4725 *) rand_win.c fix for Borland C. 4726 [Ulf M�ller] 4727 4728 *) BN_rshift bugfix for n == 0. 4729 [Bodo Moeller] 4730 4731 *) Add a 'bctest' script that checks for some known 'bc' bugs 4732 so that 'make test' does not abort just because 'bc' is broken. 4733 [Bodo Moeller] 4734 4735 *) Store verify_result within SSL_SESSION also for client side to 4736 avoid potential security hole. (Re-used sessions on the client side 4737 always resulted in verify_result==X509_V_OK, not using the original 4738 result of the server certificate verification.) 4739 [Lutz Jaenicke] 4740 4741 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 4742 SSL3_RT_APPLICATION_DATA, return 0. 4743 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 4744 [Bodo Moeller] 4745 4746 *) Fix SSL_peek: 4747 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 4748 releases, have been re-implemented by renaming the previous 4749 implementations of ssl2_read and ssl3_read to ssl2_read_internal 4750 and ssl3_read_internal, respectively, and adding 'peek' parameters 4751 to them. The new ssl[23]_{read,peek} functions are calls to 4752 ssl[23]_read_internal with the 'peek' flag set appropriately. 4753 A 'peek' parameter has also been added to ssl3_read_bytes, which 4754 does the actual work for ssl3_read_internal. 4755 [Bodo Moeller] 4756 4757 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 4758 the method-specific "init()" handler. Also clean up ex_data after 4759 calling the method-specific "finish()" handler. Previously, this was 4760 happening the other way round. 4761 [Geoff Thorpe] 4762 4763 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 4764 The previous value, 12, was not always sufficient for BN_mod_exp(). 4765 [Bodo Moeller] 4766 4767 *) Make sure that shared libraries get the internal name engine with 4768 the full version number and not just 0. This should mark the 4769 shared libraries as not backward compatible. Of course, this should 4770 be changed again when we can guarantee backward binary compatibility. 4771 [Richard Levitte] 4772 4773 *) Fix typo in get_cert_by_subject() in by_dir.c 4774 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 4775 4776 *) Rework the system to generate shared libraries: 4777 4778 - Make note of the expected extension for the shared libraries and 4779 if there is a need for symbolic links from for example libcrypto.so.0 4780 to libcrypto.so.0.9.7. There is extended info in Configure for 4781 that. 4782 4783 - Make as few rebuilds of the shared libraries as possible. 4784 4785 - Still avoid linking the OpenSSL programs with the shared libraries. 4786 4787 - When installing, install the shared libraries separately from the 4788 static ones. 4789 [Richard Levitte] 4790 4791 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 4792 4793 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 4794 and not in SSL_clear because the latter is also used by the 4795 accept/connect functions; previously, the settings made by 4796 SSL_set_read_ahead would be lost during the handshake. 4797 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 4798 4799 *) Correct util/mkdef.pl to be selective about disabled algorithms. 4800 Previously, it would create entries for disableed algorithms no 4801 matter what. 4802 [Richard Levitte] 4803 4804 *) Added several new manual pages for SSL_* function. 4805 [Lutz Jaenicke] 4806 4807 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 4808 4809 *) In ssl23_get_client_hello, generate an error message when faced 4810 with an initial SSL 3.0/TLS record that is too small to contain the 4811 first two bytes of the ClientHello message, i.e. client_version. 4812 (Note that this is a pathologic case that probably has never happened 4813 in real life.) The previous approach was to use the version number 4814 from the record header as a substitute; but our protocol choice 4815 should not depend on that one because it is not authenticated 4816 by the Finished messages. 4817 [Bodo Moeller] 4818 4819 *) More robust randomness gathering functions for Windows. 4820 [Jeffrey Altman <jaltman@columbia.edu>] 4821 4822 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 4823 not set then we don't setup the error code for issuer check errors 4824 to avoid possibly overwriting other errors which the callback does 4825 handle. If an application does set the flag then we assume it knows 4826 what it is doing and can handle the new informational codes 4827 appropriately. 4828 [Steve Henson] 4829 4830 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 4831 a general "ANY" type, as such it should be able to decode anything 4832 including tagged types. However it didn't check the class so it would 4833 wrongly interpret tagged types in the same way as their universal 4834 counterpart and unknown types were just rejected. Changed so that the 4835 tagged and unknown types are handled in the same way as a SEQUENCE: 4836 that is the encoding is stored intact. There is also a new type 4837 "V_ASN1_OTHER" which is used when the class is not universal, in this 4838 case we have no idea what the actual type is so we just lump them all 4839 together. 4840 [Steve Henson] 4841 4842 *) On VMS, stdout may very well lead to a file that is written to 4843 in a record-oriented fashion. That means that every write() will 4844 write a separate record, which will be read separately by the 4845 programs trying to read from it. This can be very confusing. 4846 4847 The solution is to put a BIO filter in the way that will buffer 4848 text until a linefeed is reached, and then write everything a 4849 line at a time, so every record written will be an actual line, 4850 not chunks of lines and not (usually doesn't happen, but I've 4851 seen it once) several lines in one record. BIO_f_linebuffer() is 4852 the answer. 4853 4854 Currently, it's a VMS-only method, because that's where it has 4855 been tested well enough. 4856 [Richard Levitte] 4857 4858 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 4859 it can return incorrect results. 4860 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 4861 but it was in 0.9.6-beta[12].) 4862 [Bodo Moeller] 4863 4864 *) Disable the check for content being present when verifying detached 4865 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 4866 include zero length content when signing messages. 4867 [Steve Henson] 4868 4869 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 4870 BIO_ctrl (for BIO pairs). 4871 [Bodo M�ller] 4872 4873 *) Add DSO method for VMS. 4874 [Richard Levitte] 4875 4876 *) Bug fix: Montgomery multiplication could produce results with the 4877 wrong sign. 4878 [Ulf M�ller] 4879 4880 *) Add RPM specification openssl.spec and modify it to build three 4881 packages. The default package contains applications, application 4882 documentation and run-time libraries. The devel package contains 4883 include files, static libraries and function documentation. The 4884 doc package contains the contents of the doc directory. The original 4885 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 4886 [Richard Levitte] 4887 4888 *) Add a large number of documentation files for many SSL routines. 4889 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 4890 4891 *) Add a configuration entry for Sony News 4. 4892 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 4893 4894 *) Don't set the two most significant bits to one when generating a 4895 random number < q in the DSA library. 4896 [Ulf M�ller] 4897 4898 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 4899 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 4900 the underlying transport is blocking) if a handshake took place. 4901 (The default behaviour is needed by applications such as s_client 4902 and s_server that use select() to determine when to use SSL_read; 4903 but for applications that know in advance when to expect data, it 4904 just makes things more complicated.) 4905 [Bodo Moeller] 4906 4907 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 4908 from EGD. 4909 [Ben Laurie] 4910 4911 *) Add a few more EBCDIC conditionals that make `req' and `x509' 4912 work better on such systems. 4913 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 4914 4915 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 4916 Update PKCS12_parse() so it copies the friendlyName and the 4917 keyid to the certificates aux info. 4918 [Steve Henson] 4919 4920 *) Fix bug in PKCS7_verify() which caused an infinite loop 4921 if there was more than one signature. 4922 [Sven Uszpelkat <su@celocom.de>] 4923 4924 *) Major change in util/mkdef.pl to include extra information 4925 about each symbol, as well as presentig variables as well 4926 as functions. This change means that there's n more need 4927 to rebuild the .num files when some algorithms are excluded. 4928 [Richard Levitte] 4929 4930 *) Allow the verify time to be set by an application, 4931 rather than always using the current time. 4932 [Steve Henson] 4933 4934 *) Phase 2 verify code reorganisation. The certificate 4935 verify code now looks up an issuer certificate by a 4936 number of criteria: subject name, authority key id 4937 and key usage. It also verifies self signed certificates 4938 by the same criteria. The main comparison function is 4939 X509_check_issued() which performs these checks. 4940 4941 Lot of changes were necessary in order to support this 4942 without completely rewriting the lookup code. 4943 4944 Authority and subject key identifier are now cached. 4945 4946 The LHASH 'certs' is X509_STORE has now been replaced 4947 by a STACK_OF(X509_OBJECT). This is mainly because an 4948 LHASH can't store or retrieve multiple objects with 4949 the same hash value. 4950 4951 As a result various functions (which were all internal 4952 use only) have changed to handle the new X509_STORE 4953 structure. This will break anything that messed round 4954 with X509_STORE internally. 4955 4956 The functions X509_STORE_add_cert() now checks for an 4957 exact match, rather than just subject name. 4958 4959 The X509_STORE API doesn't directly support the retrieval 4960 of multiple certificates matching a given criteria, however 4961 this can be worked round by performing a lookup first 4962 (which will fill the cache with candidate certificates) 4963 and then examining the cache for matches. This is probably 4964 the best we can do without throwing out X509_LOOKUP 4965 entirely (maybe later...). 4966 4967 The X509_VERIFY_CTX structure has been enhanced considerably. 4968 4969 All certificate lookup operations now go via a get_issuer() 4970 callback. Although this currently uses an X509_STORE it 4971 can be replaced by custom lookups. This is a simple way 4972 to bypass the X509_STORE hackery necessary to make this 4973 work and makes it possible to use more efficient techniques 4974 in future. A very simple version which uses a simple 4975 STACK for its trusted certificate store is also provided 4976 using X509_STORE_CTX_trusted_stack(). 4977 4978 The verify_cb() and verify() callbacks now have equivalents 4979 in the X509_STORE_CTX structure. 4980 4981 X509_STORE_CTX also has a 'flags' field which can be used 4982 to customise the verify behaviour. 4983 [Steve Henson] 4984 4985 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 4986 excludes S/MIME capabilities. 4987 [Steve Henson] 4988 4989 *) When a certificate request is read in keep a copy of the 4990 original encoding of the signed data and use it when outputing 4991 again. Signatures then use the original encoding rather than 4992 a decoded, encoded version which may cause problems if the 4993 request is improperly encoded. 4994 [Steve Henson] 4995 4996 *) For consistency with other BIO_puts implementations, call 4997 buffer_write(b, ...) directly in buffer_puts instead of calling 4998 BIO_write(b, ...). 4999 5000 In BIO_puts, increment b->num_write as in BIO_write. 5001 [Peter.Sylvester@EdelWeb.fr] 5002 5003 *) Fix BN_mul_word for the case where the word is 0. (We have to use 5004 BN_zero, we may not return a BIGNUM with an array consisting of 5005 words set to zero.) 5006 [Bodo Moeller] 5007 5008 *) Avoid calling abort() from within the library when problems are 5009 detected, except if preprocessor symbols have been defined 5010 (such as REF_CHECK, BN_DEBUG etc.). 5011 [Bodo Moeller] 5012 5013 *) New openssl application 'rsautl'. This utility can be 5014 used for low level RSA operations. DER public key 5015 BIO/fp routines also added. 5016 [Steve Henson] 5017 5018 *) New Configure entry and patches for compiling on QNX 4. 5019 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 5020 5021 *) A demo state-machine implementation was sponsored by 5022 Nuron (http://www.nuron.com/) and is now available in 5023 demos/state_machine. 5024 [Ben Laurie] 5025 5026 *) New options added to the 'dgst' utility for signature 5027 generation and verification. 5028 [Steve Henson] 5029 5030 *) Unrecognized PKCS#7 content types are now handled via a 5031 catch all ASN1_TYPE structure. This allows unsupported 5032 types to be stored as a "blob" and an application can 5033 encode and decode it manually. 5034 [Steve Henson] 5035 5036 *) Fix various signed/unsigned issues to make a_strex.c 5037 compile under VC++. 5038 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 5039 5040 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 5041 length if passed a buffer. ASN1_INTEGER_to_BN failed 5042 if passed a NULL BN and its argument was negative. 5043 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 5044 5045 *) Modification to PKCS#7 encoding routines to output definite 5046 length encoding. Since currently the whole structures are in 5047 memory there's not real point in using indefinite length 5048 constructed encoding. However if OpenSSL is compiled with 5049 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 5050 [Steve Henson] 5051 5052 *) Added BIO_vprintf() and BIO_vsnprintf(). 5053 [Richard Levitte] 5054 5055 *) Added more prefixes to parse for in the the strings written 5056 through a logging bio, to cover all the levels that are available 5057 through syslog. The prefixes are now: 5058 5059 PANIC, EMERG, EMR => LOG_EMERG 5060 ALERT, ALR => LOG_ALERT 5061 CRIT, CRI => LOG_CRIT 5062 ERROR, ERR => LOG_ERR 5063 WARNING, WARN, WAR => LOG_WARNING 5064 NOTICE, NOTE, NOT => LOG_NOTICE 5065 INFO, INF => LOG_INFO 5066 DEBUG, DBG => LOG_DEBUG 5067 5068 and as before, if none of those prefixes are present at the 5069 beginning of the string, LOG_ERR is chosen. 5070 5071 On Win32, the LOG_* levels are mapped according to this: 5072 5073 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 5074 LOG_WARNING => EVENTLOG_WARNING_TYPE 5075 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 5076 5077 [Richard Levitte] 5078 5079 *) Made it possible to reconfigure with just the configuration 5080 argument "reconf" or "reconfigure". The command line arguments 5081 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 5082 and are retrieved from there when reconfiguring. 5083 [Richard Levitte] 5084 5085 *) MD4 implemented. 5086 [Assar Westerlund <assar@sics.se>, Richard Levitte] 5087 5088 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 5089 [Richard Levitte] 5090 5091 *) The obj_dat.pl script was messing up the sorting of object 5092 names. The reason was that it compared the quoted version 5093 of strings as a result "OCSP" > "OCSP Signing" because 5094 " > SPACE. Changed script to store unquoted versions of 5095 names and add quotes on output. It was also omitting some 5096 names from the lookup table if they were given a default 5097 value (that is if SN is missing it is given the same 5098 value as LN and vice versa), these are now added on the 5099 grounds that if an object has a name we should be able to 5100 look it up. Finally added warning output when duplicate 5101 short or long names are found. 5102 [Steve Henson] 5103 5104 *) Changes needed for Tandem NSK. 5105 [Scott Uroff <scott@xypro.com>] 5106 5107 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 5108 RSA_padding_check_SSLv23(), special padding was never detected 5109 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 5110 version rollback attacks was not effective. 5111 5112 In s23_clnt.c, don't use special rollback-attack detection padding 5113 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 5114 client; similarly, in s23_srvr.c, don't do the rollback check if 5115 SSL 2.0 is the only protocol enabled in the server. 5116 [Bodo Moeller] 5117 5118 *) Make it possible to get hexdumps of unprintable data with 'openssl 5119 asn1parse'. By implication, the functions ASN1_parse_dump() and 5120 BIO_dump_indent() are added. 5121 [Richard Levitte] 5122 5123 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 5124 these print out strings and name structures based on various 5125 flags including RFC2253 support and proper handling of 5126 multibyte characters. Added options to the 'x509' utility 5127 to allow the various flags to be set. 5128 [Steve Henson] 5129 5130 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 5131 Also change the functions X509_cmp_current_time() and 5132 X509_gmtime_adj() work with an ASN1_TIME structure, 5133 this will enable certificates using GeneralizedTime in validity 5134 dates to be checked. 5135 [Steve Henson] 5136 5137 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 5138 negative public key encodings) on by default, 5139 NO_NEG_PUBKEY_BUG can be set to disable it. 5140 [Steve Henson] 5141 5142 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 5143 content octets. An i2c_ASN1_OBJECT is unnecessary because 5144 the encoding can be trivially obtained from the structure. 5145 [Steve Henson] 5146 5147 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 5148 not read locks (CRYPTO_r_[un]lock). 5149 [Bodo Moeller] 5150 5151 *) A first attempt at creating official support for shared 5152 libraries through configuration. I've kept it so the 5153 default is static libraries only, and the OpenSSL programs 5154 are always statically linked for now, but there are 5155 preparations for dynamic linking in place. 5156 This has been tested on Linux and Tru64. 5157 [Richard Levitte] 5158 5159 *) Randomness polling function for Win9x, as described in: 5160 Peter Gutmann, Software Generation of Practically Strong 5161 Random Numbers. 5162 [Ulf M�ller] 5163 5164 *) Fix so PRNG is seeded in req if using an already existing 5165 DSA key. 5166 [Steve Henson] 5167 5168 *) New options to smime application. -inform and -outform 5169 allow alternative formats for the S/MIME message including 5170 PEM and DER. The -content option allows the content to be 5171 specified separately. This should allow things like Netscape 5172 form signing output easier to verify. 5173 [Steve Henson] 5174 5175 *) Fix the ASN1 encoding of tags using the 'long form'. 5176 [Steve Henson] 5177 5178 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 5179 STRING types. These convert content octets to and from the 5180 underlying type. The actual tag and length octets are 5181 already assumed to have been read in and checked. These 5182 are needed because all other string types have virtually 5183 identical handling apart from the tag. By having versions 5184 of the ASN1 functions that just operate on content octets 5185 IMPLICIT tagging can be handled properly. It also allows 5186 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 5187 and ASN1_INTEGER are identical apart from the tag. 5188 [Steve Henson] 5189 5190 *) Change the handling of OID objects as follows: 5191 5192 - New object identifiers are inserted in objects.txt, following 5193 the syntax given in objects.README. 5194 - objects.pl is used to process obj_mac.num and create a new 5195 obj_mac.h. 5196 - obj_dat.pl is used to create a new obj_dat.h, using the data in 5197 obj_mac.h. 5198 5199 This is currently kind of a hack, and the perl code in objects.pl 5200 isn't very elegant, but it works as I intended. The simplest way 5201 to check that it worked correctly is to look in obj_dat.h and 5202 check the array nid_objs and make sure the objects haven't moved 5203 around (this is important!). Additions are OK, as well as 5204 consistent name changes. 5205 [Richard Levitte] 5206 5207 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 5208 [Bodo Moeller] 5209 5210 *) Addition of the command line parameter '-rand file' to 'openssl req'. 5211 The given file adds to whatever has already been seeded into the 5212 random pool through the RANDFILE configuration file option or 5213 environment variable, or the default random state file. 5214 [Richard Levitte] 5215 5216 *) mkstack.pl now sorts each macro group into lexical order. 5217 Previously the output order depended on the order the files 5218 appeared in the directory, resulting in needless rewriting 5219 of safestack.h . 5220 [Steve Henson] 5221 5222 *) Patches to make OpenSSL compile under Win32 again. Mostly 5223 work arounds for the VC++ problem that it treats func() as 5224 func(void). Also stripped out the parts of mkdef.pl that 5225 added extra typesafe functions: these no longer exist. 5226 [Steve Henson] 5227 5228 *) Reorganisation of the stack code. The macros are now all 5229 collected in safestack.h . Each macro is defined in terms of 5230 a "stack macro" of the form SKM_<name>(type, a, b). The 5231 DEBUG_SAFESTACK is now handled in terms of function casts, 5232 this has the advantage of retaining type safety without the 5233 use of additional functions. If DEBUG_SAFESTACK is not defined 5234 then the non typesafe macros are used instead. Also modified the 5235 mkstack.pl script to handle the new form. Needs testing to see 5236 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 5237 the default if no major problems. Similar behaviour for ASN1_SET_OF 5238 and PKCS12_STACK_OF. 5239 [Steve Henson] 5240 5241 *) When some versions of IIS use the 'NET' form of private key the 5242 key derivation algorithm is different. Normally MD5(password) is 5243 used as a 128 bit RC4 key. In the modified case 5244 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 5245 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 5246 as the old Netscape_RSA functions except they have an additional 5247 'sgckey' parameter which uses the modified algorithm. Also added 5248 an -sgckey command line option to the rsa utility. Thanks to 5249 Adrian Peck <bertie@ncipher.com> for posting details of the modified 5250 algorithm to openssl-dev. 5251 [Steve Henson] 5252 5253 *) The evp_local.h macros were using 'c.##kname' which resulted in 5254 invalid expansion on some systems (SCO 5.0.5 for example). 5255 Corrected to 'c.kname'. 5256 [Phillip Porch <root@theporch.com>] 5257 5258 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 5259 a STACK of email addresses from a certificate or request, these look 5260 in the subject name and the subject alternative name extensions and 5261 omit any duplicate addresses. 5262 [Steve Henson] 5263 5264 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 5265 This makes DSA verification about 2 % faster. 5266 [Bodo Moeller] 5267 5268 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 5269 (meaning that now 2^5 values will be precomputed, which is only 4 KB 5270 plus overhead for 1024 bit moduli). 5271 This makes exponentiations about 0.5 % faster for 1024 bit 5272 exponents (as measured by "openssl speed rsa2048"). 5273 [Bodo Moeller] 5274 5275 *) Rename memory handling macros to avoid conflicts with other 5276 software: 5277 Malloc => OPENSSL_malloc 5278 Malloc_locked => OPENSSL_malloc_locked 5279 Realloc => OPENSSL_realloc 5280 Free => OPENSSL_free 5281 [Richard Levitte] 5282 5283 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 5284 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 5285 [Bodo Moeller] 5286 5287 *) CygWin32 support. 5288 [John Jarvie <jjarvie@newsguy.com>] 5289 5290 *) The type-safe stack code has been rejigged. It is now only compiled 5291 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 5292 by default all type-specific stack functions are "#define"d back to 5293 standard stack functions. This results in more streamlined output 5294 but retains the type-safety checking possibilities of the original 5295 approach. 5296 [Geoff Thorpe] 5297 5298 *) The STACK code has been cleaned up, and certain type declarations 5299 that didn't make a lot of sense have been brought in line. This has 5300 also involved a cleanup of sorts in safestack.h to more correctly 5301 map type-safe stack functions onto their plain stack counterparts. 5302 This work has also resulted in a variety of "const"ifications of 5303 lots of the code, especially "_cmp" operations which should normally 5304 be prototyped with "const" parameters anyway. 5305 [Geoff Thorpe] 5306 5307 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 5308 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 5309 (The PRNG state consists of two parts, the large pool 'state' and 'md', 5310 where all of 'md' is used each time the PRNG is used, but 'state' 5311 is used only indexed by a cyclic counter. As entropy may not be 5312 well distributed from the beginning, 'md' is important as a 5313 chaining variable. However, the output function chains only half 5314 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 5315 all of 'md', and seeding with STATE_SIZE dummy bytes will result 5316 in all of 'state' being rewritten, with the new values depending 5317 on virtually all of 'md'. This overcomes the 80 bit limitation.) 5318 [Bodo Moeller] 5319 5320 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 5321 the handshake is continued after ssl_verify_cert_chain(); 5322 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 5323 can lead to 'unexplainable' connection aborts later. 5324 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 5325 5326 *) Major EVP API cipher revision. 5327 Add hooks for extra EVP features. This allows various cipher 5328 parameters to be set in the EVP interface. Support added for variable 5329 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 5330 setting of RC2 and RC5 parameters. 5331 5332 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 5333 ciphers. 5334 5335 Remove lots of duplicated code from the EVP library. For example *every* 5336 cipher init() function handles the 'iv' in the same way according to the 5337 cipher mode. They also all do nothing if the 'key' parameter is NULL and 5338 for CFB and OFB modes they zero ctx->num. 5339 5340 New functionality allows removal of S/MIME code RC2 hack. 5341 5342 Most of the routines have the same form and so can be declared in terms 5343 of macros. 5344 5345 By shifting this to the top level EVP_CipherInit() it can be removed from 5346 all individual ciphers. If the cipher wants to handle IVs or keys 5347 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 5348 flags. 5349 5350 Change lots of functions like EVP_EncryptUpdate() to now return a 5351 value: although software versions of the algorithms cannot fail 5352 any installed hardware versions can. 5353 [Steve Henson] 5354 5355 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 5356 this option is set, tolerate broken clients that send the negotiated 5357 protocol version number instead of the requested protocol version 5358 number. 5359 [Bodo Moeller] 5360 5361 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 5362 i.e. non-zero for export ciphersuites, zero otherwise. 5363 Previous versions had this flag inverted, inconsistent with 5364 rsa_tmp_cb (..._TMP_RSA_CB). 5365 [Bodo Moeller; problem reported by Amit Chopra] 5366 5367 *) Add missing DSA library text string. Work around for some IIS 5368 key files with invalid SEQUENCE encoding. 5369 [Steve Henson] 5370 5371 *) Add a document (doc/standards.txt) that list all kinds of standards 5372 and so on that are implemented in OpenSSL. 5373 [Richard Levitte] 5374 5375 *) Enhance c_rehash script. Old version would mishandle certificates 5376 with the same subject name hash and wouldn't handle CRLs at all. 5377 Added -fingerprint option to crl utility, to support new c_rehash 5378 features. 5379 [Steve Henson] 5380 5381 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 5382 [Ulf M�ller] 5383 5384 *) Fix for SSL server purpose checking. Server checking was 5385 rejecting certificates which had extended key usage present 5386 but no ssl client purpose. 5387 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 5388 5389 *) Make PKCS#12 code work with no password. The PKCS#12 spec 5390 is a little unclear about how a blank password is handled. 5391 Since the password in encoded as a BMPString with terminating 5392 double NULL a zero length password would end up as just the 5393 double NULL. However no password at all is different and is 5394 handled differently in the PKCS#12 key generation code. NS 5395 treats a blank password as zero length. MSIE treats it as no 5396 password on export: but it will try both on import. We now do 5397 the same: PKCS12_parse() tries zero length and no password if 5398 the password is set to "" or NULL (NULL is now a valid password: 5399 it wasn't before) as does the pkcs12 application. 5400 [Steve Henson] 5401 5402 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 5403 perror when PEM_read_bio_X509_REQ fails, the error message must 5404 be obtained from the error queue. 5405 [Bodo Moeller] 5406 5407 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 5408 it in ERR_remove_state if appropriate, and change ERR_get_state 5409 accordingly to avoid race conditions (this is necessary because 5410 thread_hash is no longer constant once set). 5411 [Bodo Moeller] 5412 5413 *) Bugfix for linux-elf makefile.one. 5414 [Ulf M�ller] 5415 5416 *) RSA_get_default_method() will now cause a default 5417 RSA_METHOD to be chosen if one doesn't exist already. 5418 Previously this was only set during a call to RSA_new() 5419 or RSA_new_method(NULL) meaning it was possible for 5420 RSA_get_default_method() to return NULL. 5421 [Geoff Thorpe] 5422 5423 *) Added native name translation to the existing DSO code 5424 that will convert (if the flag to do so is set) filenames 5425 that are sufficiently small and have no path information 5426 into a canonical native form. Eg. "blah" converted to 5427 "libblah.so" or "blah.dll" etc. 5428 [Geoff Thorpe] 5429 5430 *) New function ERR_error_string_n(e, buf, len) which is like 5431 ERR_error_string(e, buf), but writes at most 'len' bytes 5432 including the 0 terminator. For ERR_error_string_n, 'buf' 5433 may not be NULL. 5434 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 5435 5436 *) CONF library reworked to become more general. A new CONF 5437 configuration file reader "class" is implemented as well as a 5438 new functions (NCONF_*, for "New CONF") to handle it. The now 5439 old CONF_* functions are still there, but are reimplemented to 5440 work in terms of the new functions. Also, a set of functions 5441 to handle the internal storage of the configuration data is 5442 provided to make it easier to write new configuration file 5443 reader "classes" (I can definitely see something reading a 5444 configuration file in XML format, for example), called _CONF_*, 5445 or "the configuration storage API"... 5446 5447 The new configuration file reading functions are: 5448 5449 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 5450 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 5451 5452 NCONF_default, NCONF_WIN32 5453 5454 NCONF_dump_fp, NCONF_dump_bio 5455 5456 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 5457 NCONF_new creates a new CONF object. This works in the same way 5458 as other interfaces in OpenSSL, like the BIO interface. 5459 NCONF_dump_* dump the internal storage of the configuration file, 5460 which is useful for debugging. All other functions take the same 5461 arguments as the old CONF_* functions wth the exception of the 5462 first that must be a `CONF *' instead of a `LHASH *'. 5463 5464 To make it easer to use the new classes with the old CONF_* functions, 5465 the function CONF_set_default_method is provided. 5466 [Richard Levitte] 5467 5468 *) Add '-tls1' option to 'openssl ciphers', which was already 5469 mentioned in the documentation but had not been implemented. 5470 (This option is not yet really useful because even the additional 5471 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 5472 [Bodo Moeller] 5473 5474 *) Initial DSO code added into libcrypto for letting OpenSSL (and 5475 OpenSSL-based applications) load shared libraries and bind to 5476 them in a portable way. 5477 [Geoff Thorpe, with contributions from Richard Levitte] 5478 5479 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 5480 5481 *) Make sure _lrotl and _lrotr are only used with MSVC. 5482 5483 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 5484 (the default implementation of RAND_status). 5485 5486 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 5487 to '-clrext' (= clear extensions), as intended and documented. 5488 [Bodo Moeller; inconsistency pointed out by Michael Attili 5489 <attili@amaxo.com>] 5490 5491 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 5492 was larger than the MD block size. 5493 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 5494 5495 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 5496 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 5497 using the passed key: if the passed key was a private key the result 5498 of X509_print(), for example, would be to print out all the private key 5499 components. 5500 [Steve Henson] 5501 5502 *) des_quad_cksum() byte order bug fix. 5503 [Ulf M�ller, using the problem description in krb4-0.9.7, where 5504 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 5505 5506 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 5507 discouraged. 5508 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 5509 5510 *) For easily testing in shell scripts whether some command 5511 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 5512 returns with exit code 0 iff no command of the given name is available. 5513 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 5514 the output goes to stdout and nothing is printed to stderr. 5515 Additional arguments are always ignored. 5516 5517 Since for each cipher there is a command of the same name, 5518 the 'no-cipher' compilation switches can be tested this way. 5519 5520 ('openssl no-XXX' is not able to detect pseudo-commands such 5521 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 5522 [Bodo Moeller] 5523 5524 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 5525 [Bodo Moeller] 5526 5527 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 5528 is set; it will be thrown away anyway because each handshake creates 5529 its own key. 5530 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 5531 to parameters -- in previous versions (since OpenSSL 0.9.3) the 5532 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 5533 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 5534 [Bodo Moeller] 5535 5536 *) New s_client option -ign_eof: EOF at stdin is ignored, and 5537 'Q' and 'R' lose their special meanings (quit/renegotiate). 5538 This is part of what -quiet does; unlike -quiet, -ign_eof 5539 does not suppress any output. 5540 [Richard Levitte] 5541 5542 *) Add compatibility options to the purpose and trust code. The 5543 purpose X509_PURPOSE_ANY is "any purpose" which automatically 5544 accepts a certificate or CA, this was the previous behaviour, 5545 with all the associated security issues. 5546 5547 X509_TRUST_COMPAT is the old trust behaviour: only and 5548 automatically trust self signed roots in certificate store. A 5549 new trust setting X509_TRUST_DEFAULT is used to specify that 5550 a purpose has no associated trust setting and it should instead 5551 use the value in the default purpose. 5552 [Steve Henson] 5553 5554 *) Fix the PKCS#8 DSA private key code so it decodes keys again 5555 and fix a memory leak. 5556 [Steve Henson] 5557 5558 *) In util/mkerr.pl (which implements 'make errors'), preserve 5559 reason strings from the previous version of the .c file, as 5560 the default to have only downcase letters (and digits) in 5561 automatically generated reasons codes is not always appropriate. 5562 [Bodo Moeller] 5563 5564 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 5565 using strerror. Previously, ERR_reason_error_string() returned 5566 library names as reason strings for SYSerr; but SYSerr is a special 5567 case where small numbers are errno values, not library numbers. 5568 [Bodo Moeller] 5569 5570 *) Add '-dsaparam' option to 'openssl dhparam' application. This 5571 converts DSA parameters into DH parameters. (When creating parameters, 5572 DSA_generate_parameters is used.) 5573 [Bodo Moeller] 5574 5575 *) Include 'length' (recommended exponent length) in C code generated 5576 by 'openssl dhparam -C'. 5577 [Bodo Moeller] 5578 5579 *) The second argument to set_label in perlasm was already being used 5580 so couldn't be used as a "file scope" flag. Moved to third argument 5581 which was free. 5582 [Steve Henson] 5583 5584 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 5585 instead of RAND_bytes for encryption IVs and salts. 5586 [Bodo Moeller] 5587 5588 *) Include RAND_status() into RAND_METHOD instead of implementing 5589 it only for md_rand.c Otherwise replacing the PRNG by calling 5590 RAND_set_rand_method would be impossible. 5591 [Bodo Moeller] 5592 5593 *) Don't let DSA_generate_key() enter an infinite loop if the random 5594 number generation fails. 5595 [Bodo Moeller] 5596 5597 *) New 'rand' application for creating pseudo-random output. 5598 [Bodo Moeller] 5599 5600 *) Added configuration support for Linux/IA64 5601 [Rolf Haberrecker <rolf@suse.de>] 5602 5603 *) Assembler module support for Mingw32. 5604 [Ulf M�ller] 5605 5606 *) Shared library support for HPUX (in shlib/). 5607 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 5608 5609 *) Shared library support for Solaris gcc. 5610 [Lutz Behnke <behnke@trustcenter.de>] 5611 5612 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 5613 5614 *) PKCS7_encrypt() was adding text MIME headers twice because they 5615 were added manually and by SMIME_crlf_copy(). 5616 [Steve Henson] 5617 5618 *) In bntest.c don't call BN_rand with zero bits argument. 5619 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 5620 5621 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 5622 case was implemented. This caused BN_div_recp() to fail occasionally. 5623 [Ulf M�ller] 5624 5625 *) Add an optional second argument to the set_label() in the perl 5626 assembly language builder. If this argument exists and is set 5627 to 1 it signals that the assembler should use a symbol whose 5628 scope is the entire file, not just the current function. This 5629 is needed with MASM which uses the format label:: for this scope. 5630 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 5631 5632 *) Change the ASN1 types so they are typedefs by default. Before 5633 almost all types were #define'd to ASN1_STRING which was causing 5634 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 5635 for example. 5636 [Steve Henson] 5637 5638 *) Change names of new functions to the new get1/get0 naming 5639 convention: After 'get1', the caller owns a reference count 5640 and has to call ..._free; 'get0' returns a pointer to some 5641 data structure without incrementing reference counters. 5642 (Some of the existing 'get' functions increment a reference 5643 counter, some don't.) 5644 Similarly, 'set1' and 'add1' functions increase reference 5645 counters or duplicate objects. 5646 [Steve Henson] 5647 5648 *) Allow for the possibility of temp RSA key generation failure: 5649 the code used to assume it always worked and crashed on failure. 5650 [Steve Henson] 5651 5652 *) Fix potential buffer overrun problem in BIO_printf(). 5653 [Ulf M�ller, using public domain code by Patrick Powell; problem 5654 pointed out by David Sacerdote <das33@cornell.edu>] 5655 5656 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 5657 RAND_egd() and RAND_status(). In the command line application, 5658 the EGD socket can be specified like a seed file using RANDFILE 5659 or -rand. 5660 [Ulf M�ller] 5661 5662 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 5663 Some CAs (e.g. Verisign) distribute certificates in this form. 5664 [Steve Henson] 5665 5666 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 5667 list to exclude them. This means that no special compilation option 5668 is needed to use anonymous DH: it just needs to be included in the 5669 cipher list. 5670 [Steve Henson] 5671 5672 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 5673 EVP_MD_type. The old functionality is available in a new macro called 5674 EVP_MD_md(). Change code that uses it and update docs. 5675 [Steve Henson] 5676 5677 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 5678 where the 'void *' argument is replaced by a function pointer argument. 5679 Previously 'void *' was abused to point to functions, which works on 5680 many platforms, but is not correct. As these functions are usually 5681 called by macros defined in OpenSSL header files, most source code 5682 should work without changes. 5683 [Richard Levitte] 5684 5685 *) <openssl/opensslconf.h> (which is created by Configure) now contains 5686 sections with information on -D... compiler switches used for 5687 compiling the library so that applications can see them. To enable 5688 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 5689 must be defined. E.g., 5690 #define OPENSSL_ALGORITHM_DEFINES 5691 #include <openssl/opensslconf.h> 5692 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 5693 [Richard Levitte, Ulf and Bodo M�ller] 5694 5695 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 5696 record layer. 5697 [Bodo Moeller] 5698 5699 *) Change the 'other' type in certificate aux info to a STACK_OF 5700 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 5701 the required ASN1 format: arbitrary types determined by an OID. 5702 [Steve Henson] 5703 5704 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 5705 argument to 'req'. This is not because the function is newer or 5706 better than others it just uses the work 'NEW' in the certificate 5707 request header lines. Some software needs this. 5708 [Steve Henson] 5709 5710 *) Reorganise password command line arguments: now passwords can be 5711 obtained from various sources. Delete the PEM_cb function and make 5712 it the default behaviour: i.e. if the callback is NULL and the 5713 usrdata argument is not NULL interpret it as a null terminated pass 5714 phrase. If usrdata and the callback are NULL then the pass phrase 5715 is prompted for as usual. 5716 [Steve Henson] 5717 5718 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 5719 the support is automatically enabled. The resulting binaries will 5720 autodetect the card and use it if present. 5721 [Ben Laurie and Compaq Inc.] 5722 5723 *) Work around for Netscape hang bug. This sends certificate request 5724 and server done in one record. Since this is perfectly legal in the 5725 SSL/TLS protocol it isn't a "bug" option and is on by default. See 5726 the bugs/SSLv3 entry for more info. 5727 [Steve Henson] 5728 5729 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 5730 [Andy Polyakov] 5731 5732 *) Add -rand argument to smime and pkcs12 applications and read/write 5733 of seed file. 5734 [Steve Henson] 5735 5736 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 5737 [Bodo Moeller] 5738 5739 *) Add command line password options to the remaining applications. 5740 [Steve Henson] 5741 5742 *) Bug fix for BN_div_recp() for numerators with an even number of 5743 bits. 5744 [Ulf M�ller] 5745 5746 *) More tests in bntest.c, and changed test_bn output. 5747 [Ulf M�ller] 5748 5749 *) ./config recognizes MacOS X now. 5750 [Andy Polyakov] 5751 5752 *) Bug fix for BN_div() when the first words of num and divsor are 5753 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 5754 [Ulf M�ller] 5755 5756 *) Add support for various broken PKCS#8 formats, and command line 5757 options to produce them. 5758 [Steve Henson] 5759 5760 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 5761 get temporary BIGNUMs from a BN_CTX. 5762 [Ulf M�ller] 5763 5764 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 5765 for p == 0. 5766 [Ulf M�ller] 5767 5768 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 5769 include a #define from the old name to the new. The original intent 5770 was that statically linked binaries could for example just call 5771 SSLeay_add_all_ciphers() to just add ciphers to the table and not 5772 link with digests. This never worked becayse SSLeay_add_all_digests() 5773 and SSLeay_add_all_ciphers() were in the same source file so calling 5774 one would link with the other. They are now in separate source files. 5775 [Steve Henson] 5776 5777 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 5778 [Steve Henson] 5779 5780 *) Use a less unusual form of the Miller-Rabin primality test (it used 5781 a binary algorithm for exponentiation integrated into the Miller-Rabin 5782 loop, our standard modexp algorithms are faster). 5783 [Bodo Moeller] 5784 5785 *) Support for the EBCDIC character set completed. 5786 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 5787 5788 *) Source code cleanups: use const where appropriate, eliminate casts, 5789 use void * instead of char * in lhash. 5790 [Ulf M�ller] 5791 5792 *) Bugfix: ssl3_send_server_key_exchange was not restartable 5793 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 5794 this the server could overwrite ephemeral keys that the client 5795 has already seen). 5796 [Bodo Moeller] 5797 5798 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 5799 using 50 iterations of the Rabin-Miller test. 5800 5801 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 5802 iterations of the Rabin-Miller test as required by the appendix 5803 to FIPS PUB 186[-1]) instead of DSA_is_prime. 5804 As BN_is_prime_fasttest includes trial division, DSA parameter 5805 generation becomes much faster. 5806 5807 This implies a change for the callback functions in DSA_is_prime 5808 and DSA_generate_parameters: The callback function is called once 5809 for each positive witness in the Rabin-Miller test, not just 5810 occasionally in the inner loop; and the parameters to the 5811 callback function now provide an iteration count for the outer 5812 loop rather than for the current invocation of the inner loop. 5813 DSA_generate_parameters additionally can call the callback 5814 function with an 'iteration count' of -1, meaning that a 5815 candidate has passed the trial division test (when q is generated 5816 from an application-provided seed, trial division is skipped). 5817 [Bodo Moeller] 5818 5819 *) New function BN_is_prime_fasttest that optionally does trial 5820 division before starting the Rabin-Miller test and has 5821 an additional BN_CTX * argument (whereas BN_is_prime always 5822 has to allocate at least one BN_CTX). 5823 'callback(1, -1, cb_arg)' is called when a number has passed the 5824 trial division stage. 5825 [Bodo Moeller] 5826 5827 *) Fix for bug in CRL encoding. The validity dates weren't being handled 5828 as ASN1_TIME. 5829 [Steve Henson] 5830 5831 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 5832 [Steve Henson] 5833 5834 *) New function BN_pseudo_rand(). 5835 [Ulf M�ller] 5836 5837 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 5838 bignum version of BN_from_montgomery() with the working code from 5839 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 5840 the comments. 5841 [Ulf M�ller] 5842 5843 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 5844 made it impossible to use the same SSL_SESSION data structure in 5845 SSL2 clients in multiple threads. 5846 [Bodo Moeller] 5847 5848 *) The return value of RAND_load_file() no longer counts bytes obtained 5849 by stat(). RAND_load_file(..., -1) is new and uses the complete file 5850 to seed the PRNG (previously an explicit byte count was required). 5851 [Ulf M�ller, Bodo M�ller] 5852 5853 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 5854 used (char *) instead of (void *) and had casts all over the place. 5855 [Steve Henson] 5856 5857 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 5858 [Ulf M�ller] 5859 5860 *) Retain source code compatibility for BN_prime_checks macro: 5861 BN_is_prime(..., BN_prime_checks, ...) now uses 5862 BN_prime_checks_for_size to determine the appropriate number of 5863 Rabin-Miller iterations. 5864 [Ulf M�ller] 5865 5866 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 5867 DH_CHECK_P_NOT_SAFE_PRIME. 5868 (Check if this is true? OpenPGP calls them "strong".) 5869 [Ulf M�ller] 5870 5871 *) Merge the functionality of "dh" and "gendh" programs into a new program 5872 "dhparam". The old programs are retained for now but will handle DH keys 5873 (instead of parameters) in future. 5874 [Steve Henson] 5875 5876 *) Make the ciphers, s_server and s_client programs check the return values 5877 when a new cipher list is set. 5878 [Steve Henson] 5879 5880 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 5881 ciphers. Before when the 56bit ciphers were enabled the sorting was 5882 wrong. 5883 5884 The syntax for the cipher sorting has been extended to support sorting by 5885 cipher-strength (using the strength_bits hard coded in the tables). 5886 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 5887 5888 Fix a bug in the cipher-command parser: when supplying a cipher command 5889 string with an "undefined" symbol (neither command nor alphanumeric 5890 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 5891 an error is flagged. 5892 5893 Due to the strength-sorting extension, the code of the 5894 ssl_create_cipher_list() function was completely rearranged. I hope that 5895 the readability was also increased :-) 5896 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5897 5898 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 5899 for the first serial number and places 2 in the serial number file. This 5900 avoids problems when the root CA is created with serial number zero and 5901 the first user certificate has the same issuer name and serial number 5902 as the root CA. 5903 [Steve Henson] 5904 5905 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 5906 the new code. Add documentation for this stuff. 5907 [Steve Henson] 5908 5909 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 5910 X509_*() to X509at_*() on the grounds that they don't handle X509 5911 structures and behave in an analagous way to the X509v3 functions: 5912 they shouldn't be called directly but wrapper functions should be used 5913 instead. 5914 5915 So we also now have some wrapper functions that call the X509at functions 5916 when passed certificate requests. (TO DO: similar things can be done with 5917 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 5918 things. Some of these need some d2i or i2d and print functionality 5919 because they handle more complex structures.) 5920 [Steve Henson] 5921 5922 *) Add missing #ifndefs that caused missing symbols when building libssl 5923 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 5924 NO_RSA in ssl/s2*.c. 5925 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 5926 5927 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 5928 has a return value which indicates the quality of the random data 5929 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 5930 error queue. New function RAND_pseudo_bytes() generates output that is 5931 guaranteed to be unique but not unpredictable. RAND_add is like 5932 RAND_seed, but takes an extra argument for an entropy estimate 5933 (RAND_seed always assumes full entropy). 5934 [Ulf M�ller] 5935 5936 *) Do more iterations of Rabin-Miller probable prime test (specifically, 5937 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 5938 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 5939 in crypto/bn/bn_prime.c for the complete table). This guarantees a 5940 false-positive rate of at most 2^-80 for random input. 5941 [Bodo Moeller] 5942 5943 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 5944 [Bodo Moeller] 5945 5946 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 5947 in the 0.9.5 release), this returns the chain 5948 from an X509_CTX structure with a dup of the stack and all 5949 the X509 reference counts upped: so the stack will exist 5950 after X509_CTX_cleanup() has been called. Modify pkcs12.c 5951 to use this. 5952 5953 Also make SSL_SESSION_print() print out the verify return 5954 code. 5955 [Steve Henson] 5956 5957 *) Add manpage for the pkcs12 command. Also change the default 5958 behaviour so MAC iteration counts are used unless the new 5959 -nomaciter option is used. This improves file security and 5960 only older versions of MSIE (4.0 for example) need it. 5961 [Steve Henson] 5962 5963 *) Honor the no-xxx Configure options when creating .DEF files. 5964 [Ulf M�ller] 5965 5966 *) Add PKCS#10 attributes to field table: challengePassword, 5967 unstructuredName and unstructuredAddress. These are taken from 5968 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 5969 international characters are used. 5970 5971 More changes to X509_ATTRIBUTE code: allow the setting of types 5972 based on strings. Remove the 'loc' parameter when adding 5973 attributes because these will be a SET OF encoding which is sorted 5974 in ASN1 order. 5975 [Steve Henson] 5976 5977 *) Initial changes to the 'req' utility to allow request generation 5978 automation. This will allow an application to just generate a template 5979 file containing all the field values and have req construct the 5980 request. 5981 5982 Initial support for X509_ATTRIBUTE handling. Stacks of these are 5983 used all over the place including certificate requests and PKCS#7 5984 structures. They are currently handled manually where necessary with 5985 some primitive wrappers for PKCS#7. The new functions behave in a 5986 manner analogous to the X509 extension functions: they allow 5987 attributes to be looked up by NID and added. 5988 5989 Later something similar to the X509V3 code would be desirable to 5990 automatically handle the encoding, decoding and printing of the 5991 more complex types. The string types like challengePassword can 5992 be handled by the string table functions. 5993 5994 Also modified the multi byte string table handling. Now there is 5995 a 'global mask' which masks out certain types. The table itself 5996 can use the flag STABLE_NO_MASK to ignore the mask setting: this 5997 is useful when for example there is only one permissible type 5998 (as in countryName) and using the mask might result in no valid 5999 types at all. 6000 [Steve Henson] 6001 6002 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 6003 SSL_get_peer_finished to allow applications to obtain the latest 6004 Finished messages sent to the peer or expected from the peer, 6005 respectively. (SSL_get_peer_finished is usually the Finished message 6006 actually received from the peer, otherwise the protocol will be aborted.) 6007 6008 As the Finished message are message digests of the complete handshake 6009 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 6010 be used for external authentication procedures when the authentication 6011 provided by SSL/TLS is not desired or is not enough. 6012 [Bodo Moeller] 6013 6014 *) Enhanced support for Alpha Linux is added. Now ./config checks if 6015 the host supports BWX extension and if Compaq C is present on the 6016 $PATH. Just exploiting of the BWX extension results in 20-30% 6017 performance kick for some algorithms, e.g. DES and RC4 to mention 6018 a couple. Compaq C in turn generates ~20% faster code for MD5 and 6019 SHA1. 6020 [Andy Polyakov] 6021 6022 *) Add support for MS "fast SGC". This is arguably a violation of the 6023 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 6024 weak crypto and after checking the certificate is SGC a second one 6025 with strong crypto. MS SGC stops the first handshake after receiving 6026 the server certificate message and sends a second client hello. Since 6027 a server will typically do all the time consuming operations before 6028 expecting any further messages from the client (server key exchange 6029 is the most expensive) there is little difference between the two. 6030 6031 To get OpenSSL to support MS SGC we have to permit a second client 6032 hello message after we have sent server done. In addition we have to 6033 reset the MAC if we do get this second client hello. 6034 [Steve Henson] 6035 6036 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 6037 if a DER encoded private key is RSA or DSA traditional format. Changed 6038 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 6039 format DER encoded private key. Newer code should use PKCS#8 format which 6040 has the key type encoded in the ASN1 structure. Added DER private key 6041 support to pkcs8 application. 6042 [Steve Henson] 6043 6044 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 6045 ciphersuites has been selected (as required by the SSL 3/TLS 1 6046 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 6047 is set, we interpret this as a request to violate the specification 6048 (the worst that can happen is a handshake failure, and 'correct' 6049 behaviour would result in a handshake failure anyway). 6050 [Bodo Moeller] 6051 6052 *) In SSL_CTX_add_session, take into account that there might be multiple 6053 SSL_SESSION structures with the same session ID (e.g. when two threads 6054 concurrently obtain them from an external cache). 6055 The internal cache can handle only one SSL_SESSION with a given ID, 6056 so if there's a conflict, we now throw out the old one to achieve 6057 consistency. 6058 [Bodo Moeller] 6059 6060 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 6061 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 6062 some routines that use cipher OIDs: some ciphers do not have OIDs 6063 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 6064 example. 6065 [Steve Henson] 6066 6067 *) Simplify the trust setting structure and code. Now we just have 6068 two sequences of OIDs for trusted and rejected settings. These will 6069 typically have values the same as the extended key usage extension 6070 and any application specific purposes. 6071 6072 The trust checking code now has a default behaviour: it will just 6073 check for an object with the same NID as the passed id. Functions can 6074 be provided to override either the default behaviour or the behaviour 6075 for a given id. SSL client, server and email already have functions 6076 in place for compatibility: they check the NID and also return "trusted" 6077 if the certificate is self signed. 6078 [Steve Henson] 6079 6080 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 6081 traditional format into an EVP_PKEY structure. 6082 [Steve Henson] 6083 6084 *) Add a password callback function PEM_cb() which either prompts for 6085 a password if usr_data is NULL or otherwise assumes it is a null 6086 terminated password. Allow passwords to be passed on command line 6087 environment or config files in a few more utilities. 6088 [Steve Henson] 6089 6090 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 6091 keys. Add some short names for PKCS#8 PBE algorithms and allow them 6092 to be specified on the command line for the pkcs8 and pkcs12 utilities. 6093 Update documentation. 6094 [Steve Henson] 6095 6096 *) Support for ASN1 "NULL" type. This could be handled before by using 6097 ASN1_TYPE but there wasn't any function that would try to read a NULL 6098 and produce an error if it couldn't. For compatibility we also have 6099 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 6100 don't allocate anything because they don't need to. 6101 [Steve Henson] 6102 6103 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 6104 for details. 6105 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 6106 6107 *) Rebuild of the memory allocation routines used by OpenSSL code and 6108 possibly others as well. The purpose is to make an interface that 6109 provide hooks so anyone can build a separate set of allocation and 6110 deallocation routines to be used by OpenSSL, for example memory 6111 pool implementations, or something else, which was previously hard 6112 since Malloc(), Realloc() and Free() were defined as macros having 6113 the values malloc, realloc and free, respectively (except for Win32 6114 compilations). The same is provided for memory debugging code. 6115 OpenSSL already comes with functionality to find memory leaks, but 6116 this gives people a chance to debug other memory problems. 6117 6118 With these changes, a new set of functions and macros have appeared: 6119 6120 CRYPTO_set_mem_debug_functions() [F] 6121 CRYPTO_get_mem_debug_functions() [F] 6122 CRYPTO_dbg_set_options() [F] 6123 CRYPTO_dbg_get_options() [F] 6124 CRYPTO_malloc_debug_init() [M] 6125 6126 The memory debug functions are NULL by default, unless the library 6127 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 6128 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 6129 gives the standard debugging functions that come with OpenSSL) or 6130 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 6131 provided by the library user) must be used. When the standard 6132 debugging functions are used, CRYPTO_dbg_set_options can be used to 6133 request additional information: 6134 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 6135 the CRYPTO_MDEBUG_xxx macro when compiling the library. 6136 6137 Also, things like CRYPTO_set_mem_functions will always give the 6138 expected result (the new set of functions is used for allocation 6139 and deallocation) at all times, regardless of platform and compiler 6140 options. 6141 6142 To finish it up, some functions that were never use in any other 6143 way than through macros have a new API and new semantic: 6144 6145 CRYPTO_dbg_malloc() 6146 CRYPTO_dbg_realloc() 6147 CRYPTO_dbg_free() 6148 6149 All macros of value have retained their old syntax. 6150 [Richard Levitte and Bodo Moeller] 6151 6152 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 6153 ordering of SMIMECapabilities wasn't in "strength order" and there 6154 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 6155 algorithm. 6156 [Steve Henson] 6157 6158 *) Some ASN1 types with illegal zero length encoding (INTEGER, 6159 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 6160 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 6161 6162 *) Merge in my S/MIME library for OpenSSL. This provides a simple 6163 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 6164 functionality to handle multipart/signed properly) and a utility 6165 called 'smime' to call all this stuff. This is based on code I 6166 originally wrote for Celo who have kindly allowed it to be 6167 included in OpenSSL. 6168 [Steve Henson] 6169 6170 *) Add variants des_set_key_checked and des_set_key_unchecked of 6171 des_set_key (aka des_key_sched). Global variable des_check_key 6172 decides which of these is called by des_set_key; this way 6173 des_check_key behaves as it always did, but applications and 6174 the library itself, which was buggy for des_check_key == 1, 6175 have a cleaner way to pick the version they need. 6176 [Bodo Moeller] 6177 6178 *) New function PKCS12_newpass() which changes the password of a 6179 PKCS12 structure. 6180 [Steve Henson] 6181 6182 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 6183 dynamic mix. In both cases the ids can be used as an index into the 6184 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 6185 functions so they accept a list of the field values and the 6186 application doesn't need to directly manipulate the X509_TRUST 6187 structure. 6188 [Steve Henson] 6189 6190 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 6191 need initialising. 6192 [Steve Henson] 6193 6194 *) Modify the way the V3 extension code looks up extensions. This now 6195 works in a similar way to the object code: we have some "standard" 6196 extensions in a static table which is searched with OBJ_bsearch() 6197 and the application can add dynamic ones if needed. The file 6198 crypto/x509v3/ext_dat.h now has the info: this file needs to be 6199 updated whenever a new extension is added to the core code and kept 6200 in ext_nid order. There is a simple program 'tabtest.c' which checks 6201 this. New extensions are not added too often so this file can readily 6202 be maintained manually. 6203 6204 There are two big advantages in doing things this way. The extensions 6205 can be looked up immediately and no longer need to be "added" using 6206 X509V3_add_standard_extensions(): this function now does nothing. 6207 [Side note: I get *lots* of email saying the extension code doesn't 6208 work because people forget to call this function] 6209 Also no dynamic allocation is done unless new extensions are added: 6210 so if we don't add custom extensions there is no need to call 6211 X509V3_EXT_cleanup(). 6212 [Steve Henson] 6213 6214 *) Modify enc utility's salting as follows: make salting the default. Add a 6215 magic header, so unsalted files fail gracefully instead of just decrypting 6216 to garbage. This is because not salting is a big security hole, so people 6217 should be discouraged from doing it. 6218 [Ben Laurie] 6219 6220 *) Fixes and enhancements to the 'x509' utility. It allowed a message 6221 digest to be passed on the command line but it only used this 6222 parameter when signing a certificate. Modified so all relevant 6223 operations are affected by the digest parameter including the 6224 -fingerprint and -x509toreq options. Also -x509toreq choked if a 6225 DSA key was used because it didn't fix the digest. 6226 [Steve Henson] 6227 6228 *) Initial certificate chain verify code. Currently tests the untrusted 6229 certificates for consistency with the verify purpose (which is set 6230 when the X509_STORE_CTX structure is set up) and checks the pathlength. 6231 6232 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 6233 this is because it will reject chains with invalid extensions whereas 6234 every previous version of OpenSSL and SSLeay made no checks at all. 6235 6236 Trust code: checks the root CA for the relevant trust settings. Trust 6237 settings have an initial value consistent with the verify purpose: e.g. 6238 if the verify purpose is for SSL client use it expects the CA to be 6239 trusted for SSL client use. However the default value can be changed to 6240 permit custom trust settings: one example of this would be to only trust 6241 certificates from a specific "secure" set of CAs. 6242 6243 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 6244 which should be used for version portability: especially since the 6245 verify structure is likely to change more often now. 6246 6247 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 6248 to set them. If not set then assume SSL clients will verify SSL servers 6249 and vice versa. 6250 6251 Two new options to the verify program: -untrusted allows a set of 6252 untrusted certificates to be passed in and -purpose which sets the 6253 intended purpose of the certificate. If a purpose is set then the 6254 new chain verify code is used to check extension consistency. 6255 [Steve Henson] 6256 6257 *) Support for the authority information access extension. 6258 [Steve Henson] 6259 6260 *) Modify RSA and DSA PEM read routines to transparently handle 6261 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 6262 public keys in a format compatible with certificate 6263 SubjectPublicKeyInfo structures. Unfortunately there were already 6264 functions called *_PublicKey_* which used various odd formats so 6265 these are retained for compatibility: however the DSA variants were 6266 never in a public release so they have been deleted. Changed dsa/rsa 6267 utilities to handle the new format: note no releases ever handled public 6268 keys so we should be OK. 6269 6270 The primary motivation for this change is to avoid the same fiasco 6271 that dogs private keys: there are several incompatible private key 6272 formats some of which are standard and some OpenSSL specific and 6273 require various evil hacks to allow partial transparent handling and 6274 even then it doesn't work with DER formats. Given the option anything 6275 other than PKCS#8 should be dumped: but the other formats have to 6276 stay in the name of compatibility. 6277 6278 With public keys and the benefit of hindsight one standard format 6279 is used which works with EVP_PKEY, RSA or DSA structures: though 6280 it clearly returns an error if you try to read the wrong kind of key. 6281 6282 Added a -pubkey option to the 'x509' utility to output the public key. 6283 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 6284 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 6285 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 6286 that do the same as the EVP_PKEY_assign_*() except they up the 6287 reference count of the added key (they don't "swallow" the 6288 supplied key). 6289 [Steve Henson] 6290 6291 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 6292 CRLs would fail if the file contained no certificates or no CRLs: 6293 added a new function to read in both types and return the number 6294 read: this means that if none are read it will be an error. The 6295 DER versions of the certificate and CRL reader would always fail 6296 because it isn't possible to mix certificates and CRLs in DER format 6297 without choking one or the other routine. Changed this to just read 6298 a certificate: this is the best we can do. Also modified the code 6299 in apps/verify.c to take notice of return codes: it was previously 6300 attempting to read in certificates from NULL pointers and ignoring 6301 any errors: this is one reason why the cert and CRL reader seemed 6302 to work. It doesn't check return codes from the default certificate 6303 routines: these may well fail if the certificates aren't installed. 6304 [Steve Henson] 6305 6306 *) Code to support otherName option in GeneralName. 6307 [Steve Henson] 6308 6309 *) First update to verify code. Change the verify utility 6310 so it warns if it is passed a self signed certificate: 6311 for consistency with the normal behaviour. X509_verify 6312 has been modified to it will now verify a self signed 6313 certificate if *exactly* the same certificate appears 6314 in the store: it was previously impossible to trust a 6315 single self signed certificate. This means that: 6316 openssl verify ss.pem 6317 now gives a warning about a self signed certificate but 6318 openssl verify -CAfile ss.pem ss.pem 6319 is OK. 6320 [Steve Henson] 6321 6322 *) For servers, store verify_result in SSL_SESSION data structure 6323 (and add it to external session representation). 6324 This is needed when client certificate verifications fails, 6325 but an application-provided verification callback (set by 6326 SSL_CTX_set_cert_verify_callback) allows accepting the session 6327 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 6328 but returns 1): When the session is reused, we have to set 6329 ssl->verify_result to the appropriate error code to avoid 6330 security holes. 6331 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 6332 6333 *) Fix a bug in the new PKCS#7 code: it didn't consider the 6334 case in PKCS7_dataInit() where the signed PKCS7 structure 6335 didn't contain any existing data because it was being created. 6336 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 6337 6338 *) Add a salt to the key derivation routines in enc.c. This 6339 forms the first 8 bytes of the encrypted file. Also add a 6340 -S option to allow a salt to be input on the command line. 6341 [Steve Henson] 6342 6343 *) New function X509_cmp(). Oddly enough there wasn't a function 6344 to compare two certificates. We do this by working out the SHA1 6345 hash and comparing that. X509_cmp() will be needed by the trust 6346 code. 6347 [Steve Henson] 6348 6349 *) SSL_get1_session() is like SSL_get_session(), but increments 6350 the reference count in the SSL_SESSION returned. 6351 [Geoff Thorpe <geoff@eu.c2.net>] 6352 6353 *) Fix for 'req': it was adding a null to request attributes. 6354 Also change the X509_LOOKUP and X509_INFO code to handle 6355 certificate auxiliary information. 6356 [Steve Henson] 6357 6358 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 6359 the 'enc' command. 6360 [Steve Henson] 6361 6362 *) Add the possibility to add extra information to the memory leak 6363 detecting output, to form tracebacks, showing from where each 6364 allocation was originated: CRYPTO_push_info("constant string") adds 6365 the string plus current file name and line number to a per-thread 6366 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 6367 is like calling CYRPTO_pop_info() until the stack is empty. 6368 Also updated memory leak detection code to be multi-thread-safe. 6369 [Richard Levitte] 6370 6371 *) Add options -text and -noout to pkcs7 utility and delete the 6372 encryption options which never did anything. Update docs. 6373 [Steve Henson] 6374 6375 *) Add options to some of the utilities to allow the pass phrase 6376 to be included on either the command line (not recommended on 6377 OSes like Unix) or read from the environment. Update the 6378 manpages and fix a few bugs. 6379 [Steve Henson] 6380 6381 *) Add a few manpages for some of the openssl commands. 6382 [Steve Henson] 6383 6384 *) Fix the -revoke option in ca. It was freeing up memory twice, 6385 leaking and not finding already revoked certificates. 6386 [Steve Henson] 6387 6388 *) Extensive changes to support certificate auxiliary information. 6389 This involves the use of X509_CERT_AUX structure and X509_AUX 6390 functions. An X509_AUX function such as PEM_read_X509_AUX() 6391 can still read in a certificate file in the usual way but it 6392 will also read in any additional "auxiliary information". By 6393 doing things this way a fair degree of compatibility can be 6394 retained: existing certificates can have this information added 6395 using the new 'x509' options. 6396 6397 Current auxiliary information includes an "alias" and some trust 6398 settings. The trust settings will ultimately be used in enhanced 6399 certificate chain verification routines: currently a certificate 6400 can only be trusted if it is self signed and then it is trusted 6401 for all purposes. 6402 [Steve Henson] 6403 6404 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 6405 The problem was that one of the replacement routines had not been working 6406 since SSLeay releases. For now the offending routine has been replaced 6407 with non-optimised assembler. Even so, this now gives around 95% 6408 performance improvement for 1024 bit RSA signs. 6409 [Mark Cox] 6410 6411 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 6412 handling. Most clients have the effective key size in bits equal to 6413 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 6414 A few however don't do this and instead use the size of the decrypted key 6415 to determine the RC2 key length and the AlgorithmIdentifier to determine 6416 the effective key length. In this case the effective key length can still 6417 be 40 bits but the key length can be 168 bits for example. This is fixed 6418 by manually forcing an RC2 key into the EVP_PKEY structure because the 6419 EVP code can't currently handle unusual RC2 key sizes: it always assumes 6420 the key length and effective key length are equal. 6421 [Steve Henson] 6422 6423 *) Add a bunch of functions that should simplify the creation of 6424 X509_NAME structures. Now you should be able to do: 6425 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 6426 and have it automatically work out the correct field type and fill in 6427 the structures. The more adventurous can try: 6428 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 6429 and it will (hopefully) work out the correct multibyte encoding. 6430 [Steve Henson] 6431 6432 *) Change the 'req' utility to use the new field handling and multibyte 6433 copy routines. Before the DN field creation was handled in an ad hoc 6434 way in req, ca, and x509 which was rather broken and didn't support 6435 BMPStrings or UTF8Strings. Since some software doesn't implement 6436 BMPStrings or UTF8Strings yet, they can be enabled using the config file 6437 using the dirstring_type option. See the new comment in the default 6438 openssl.cnf for more info. 6439 [Steve Henson] 6440 6441 *) Make crypto/rand/md_rand.c more robust: 6442 - Assure unique random numbers after fork(). 6443 - Make sure that concurrent threads access the global counter and 6444 md serializably so that we never lose entropy in them 6445 or use exactly the same state in multiple threads. 6446 Access to the large state is not always serializable because 6447 the additional locking could be a performance killer, and 6448 md should be large enough anyway. 6449 [Bodo Moeller] 6450 6451 *) New file apps/app_rand.c with commonly needed functionality 6452 for handling the random seed file. 6453 6454 Use the random seed file in some applications that previously did not: 6455 ca, 6456 dsaparam -genkey (which also ignored its '-rand' option), 6457 s_client, 6458 s_server, 6459 x509 (when signing). 6460 Except on systems with /dev/urandom, it is crucial to have a random 6461 seed file at least for key creation, DSA signing, and for DH exchanges; 6462 for RSA signatures we could do without one. 6463 6464 gendh and gendsa (unlike genrsa) used to read only the first byte 6465 of each file listed in the '-rand' option. The function as previously 6466 found in genrsa is now in app_rand.c and is used by all programs 6467 that support '-rand'. 6468 [Bodo Moeller] 6469 6470 *) In RAND_write_file, use mode 0600 for creating files; 6471 don't just chmod when it may be too late. 6472 [Bodo Moeller] 6473 6474 *) Report an error from X509_STORE_load_locations 6475 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 6476 [Bill Perry] 6477 6478 *) New function ASN1_mbstring_copy() this copies a string in either 6479 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 6480 into an ASN1_STRING type. A mask of permissible types is passed 6481 and it chooses the "minimal" type to use or an error if not type 6482 is suitable. 6483 [Steve Henson] 6484 6485 *) Add function equivalents to the various macros in asn1.h. The old 6486 macros are retained with an M_ prefix. Code inside the library can 6487 use the M_ macros. External code (including the openssl utility) 6488 should *NOT* in order to be "shared library friendly". 6489 [Steve Henson] 6490 6491 *) Add various functions that can check a certificate's extensions 6492 to see if it usable for various purposes such as SSL client, 6493 server or S/MIME and CAs of these types. This is currently 6494 VERY EXPERIMENTAL but will ultimately be used for certificate chain 6495 verification. Also added a -purpose flag to x509 utility to 6496 print out all the purposes. 6497 [Steve Henson] 6498 6499 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 6500 functions. 6501 [Steve Henson] 6502 6503 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 6504 for, obtain and decode and extension and obtain its critical flag. 6505 This allows all the necessary extension code to be handled in a 6506 single function call. 6507 [Steve Henson] 6508 6509 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 6510 platforms. See crypto/rc4/rc4_enc.c for further details. 6511 [Andy Polyakov] 6512 6513 *) New -noout option to asn1parse. This causes no output to be produced 6514 its main use is when combined with -strparse and -out to extract data 6515 from a file (which may not be in ASN.1 format). 6516 [Steve Henson] 6517 6518 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 6519 when producing the local key id. 6520 [Richard Levitte <levitte@stacken.kth.se>] 6521 6522 *) New option -dhparam in s_server. This allows a DH parameter file to be 6523 stated explicitly. If it is not stated then it tries the first server 6524 certificate file. The previous behaviour hard coded the filename 6525 "server.pem". 6526 [Steve Henson] 6527 6528 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 6529 a public key to be input or output. For example: 6530 openssl rsa -in key.pem -pubout -out pubkey.pem 6531 Also added necessary DSA public key functions to handle this. 6532 [Steve Henson] 6533 6534 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 6535 in the message. This was handled by allowing 6536 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 6537 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 6538 6539 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 6540 to the end of the strings whereas this didn't. This would cause problems 6541 if strings read with d2i_ASN1_bytes() were later modified. 6542 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 6543 6544 *) Fix for base64 decode bug. When a base64 bio reads only one line of 6545 data and it contains EOF it will end up returning an error. This is 6546 caused by input 46 bytes long. The cause is due to the way base64 6547 BIOs find the start of base64 encoded data. They do this by trying a 6548 trial decode on each line until they find one that works. When they 6549 do a flag is set and it starts again knowing it can pass all the 6550 data directly through the decoder. Unfortunately it doesn't reset 6551 the context it uses. This means that if EOF is reached an attempt 6552 is made to pass two EOFs through the context and this causes the 6553 resulting error. This can also cause other problems as well. As is 6554 usual with these problems it takes *ages* to find and the fix is 6555 trivial: move one line. 6556 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 6557 6558 *) Ugly workaround to get s_client and s_server working under Windows. The 6559 old code wouldn't work because it needed to select() on sockets and the 6560 tty (for keypresses and to see if data could be written). Win32 only 6561 supports select() on sockets so we select() with a 1s timeout on the 6562 sockets and then see if any characters are waiting to be read, if none 6563 are present then we retry, we also assume we can always write data to 6564 the tty. This isn't nice because the code then blocks until we've 6565 received a complete line of data and it is effectively polling the 6566 keyboard at 1s intervals: however it's quite a bit better than not 6567 working at all :-) A dedicated Windows application might handle this 6568 with an event loop for example. 6569 [Steve Henson] 6570 6571 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 6572 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 6573 will be called when RSA_sign() and RSA_verify() are used. This is useful 6574 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 6575 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 6576 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 6577 This necessitated the support of an extra signature type NID_md5_sha1 6578 for SSL signatures and modifications to the SSL library to use it instead 6579 of calling RSA_public_decrypt() and RSA_private_encrypt(). 6580 [Steve Henson] 6581 6582 *) Add new -verify -CAfile and -CApath options to the crl program, these 6583 will lookup a CRL issuers certificate and verify the signature in a 6584 similar way to the verify program. Tidy up the crl program so it 6585 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 6586 less strict. It will now permit CRL extensions even if it is not 6587 a V2 CRL: this will allow it to tolerate some broken CRLs. 6588 [Steve Henson] 6589 6590 *) Initialize all non-automatic variables each time one of the openssl 6591 sub-programs is started (this is necessary as they may be started 6592 multiple times from the "OpenSSL>" prompt). 6593 [Lennart Bang, Bodo Moeller] 6594 6595 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 6596 removing all other RSA functionality (this is what NO_RSA does). This 6597 is so (for example) those in the US can disable those operations covered 6598 by the RSA patent while allowing storage and parsing of RSA keys and RSA 6599 key generation. 6600 [Steve Henson] 6601 6602 *) Non-copying interface to BIO pairs. 6603 (still largely untested) 6604 [Bodo Moeller] 6605 6606 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 6607 ASCII string. This was handled independently in various places before. 6608 [Steve Henson] 6609 6610 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 6611 UTF8 strings a character at a time. 6612 [Steve Henson] 6613 6614 *) Use client_version from client hello to select the protocol 6615 (s23_srvr.c) and for RSA client key exchange verification 6616 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 6617 [Bodo Moeller] 6618 6619 *) Add various utility functions to handle SPKACs, these were previously 6620 handled by poking round in the structure internals. Added new function 6621 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 6622 print, verify and generate SPKACs. Based on an original idea from 6623 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 6624 [Steve Henson] 6625 6626 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 6627 [Andy Polyakov] 6628 6629 *) Allow the config file extension section to be overwritten on the 6630 command line. Based on an original idea from Massimiliano Pala 6631 <madwolf@comune.modena.it>. The new option is called -extensions 6632 and can be applied to ca, req and x509. Also -reqexts to override 6633 the request extensions in req and -crlexts to override the crl extensions 6634 in ca. 6635 [Steve Henson] 6636 6637 *) Add new feature to the SPKAC handling in ca. Now you can include 6638 the same field multiple times by preceding it by "XXXX." for example: 6639 1.OU="Unit name 1" 6640 2.OU="Unit name 2" 6641 this is the same syntax as used in the req config file. 6642 [Steve Henson] 6643 6644 *) Allow certificate extensions to be added to certificate requests. These 6645 are specified in a 'req_extensions' option of the req section of the 6646 config file. They can be printed out with the -text option to req but 6647 are otherwise ignored at present. 6648 [Steve Henson] 6649 6650 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 6651 data read consists of only the final block it would not decrypted because 6652 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 6653 A misplaced 'break' also meant the decrypted final block might not be 6654 copied until the next read. 6655 [Steve Henson] 6656 6657 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 6658 a few extra parameters to the DH structure: these will be useful if 6659 for example we want the value of 'q' or implement X9.42 DH. 6660 [Steve Henson] 6661 6662 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 6663 provides hooks that allow the default DSA functions or functions on a 6664 "per key" basis to be replaced. This allows hardware acceleration and 6665 hardware key storage to be handled without major modification to the 6666 library. Also added low level modexp hooks and CRYPTO_EX structure and 6667 associated functions. 6668 [Steve Henson] 6669 6670 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 6671 as "read only": it can't be written to and the buffer it points to will 6672 not be freed. Reading from a read only BIO is much more efficient than 6673 a normal memory BIO. This was added because there are several times when 6674 an area of memory needs to be read from a BIO. The previous method was 6675 to create a memory BIO and write the data to it, this results in two 6676 copies of the data and an O(n^2) reading algorithm. There is a new 6677 function BIO_new_mem_buf() which creates a read only memory BIO from 6678 an area of memory. Also modified the PKCS#7 routines to use read only 6679 memory BIOs. 6680 [Steve Henson] 6681 6682 *) Bugfix: ssl23_get_client_hello did not work properly when called in 6683 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 6684 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 6685 but a retry condition occured while trying to read the rest. 6686 [Bodo Moeller] 6687 6688 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 6689 NID_pkcs7_encrypted by default: this was wrong since this should almost 6690 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 6691 the encrypted data type: this is a more sensible place to put it and it 6692 allows the PKCS#12 code to be tidied up that duplicated this 6693 functionality. 6694 [Steve Henson] 6695 6696 *) Changed obj_dat.pl script so it takes its input and output files on 6697 the command line. This should avoid shell escape redirection problems 6698 under Win32. 6699 [Steve Henson] 6700 6701 *) Initial support for certificate extension requests, these are included 6702 in things like Xenroll certificate requests. Included functions to allow 6703 extensions to be obtained and added. 6704 [Steve Henson] 6705 6706 *) -crlf option to s_client and s_server for sending newlines as 6707 CRLF (as required by many protocols). 6708 [Bodo Moeller] 6709 6710 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 6711 6712 *) Install libRSAglue.a when OpenSSL is built with RSAref. 6713 [Ralf S. Engelschall] 6714 6715 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 6716 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 6717 6718 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 6719 program. 6720 [Steve Henson] 6721 6722 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 6723 DH parameters/keys (q is lost during that conversion, but the resulting 6724 DH parameters contain its length). 6725 6726 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 6727 much faster than DH_generate_parameters (which creates parameters 6728 where p = 2*q + 1), and also the smaller q makes DH computations 6729 much more efficient (160-bit exponentiation instead of 1024-bit 6730 exponentiation); so this provides a convenient way to support DHE 6731 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 6732 utter importance to use 6733 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6734 or 6735 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6736 when such DH parameters are used, because otherwise small subgroup 6737 attacks may become possible! 6738 [Bodo Moeller] 6739 6740 *) Avoid memory leak in i2d_DHparams. 6741 [Bodo Moeller] 6742 6743 *) Allow the -k option to be used more than once in the enc program: 6744 this allows the same encrypted message to be read by multiple recipients. 6745 [Steve Henson] 6746 6747 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 6748 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 6749 it will always use the numerical form of the OID, even if it has a short 6750 or long name. 6751 [Steve Henson] 6752 6753 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 6754 method only got called if p,q,dmp1,dmq1,iqmp components were present, 6755 otherwise bn_mod_exp was called. In the case of hardware keys for example 6756 no private key components need be present and it might store extra data 6757 in the RSA structure, which cannot be accessed from bn_mod_exp. 6758 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 6759 private key operations. 6760 [Steve Henson] 6761 6762 *) Added support for SPARC Linux. 6763 [Andy Polyakov] 6764 6765 *) pem_password_cb function type incompatibly changed from 6766 typedef int pem_password_cb(char *buf, int size, int rwflag); 6767 to 6768 ....(char *buf, int size, int rwflag, void *userdata); 6769 so that applications can pass data to their callbacks: 6770 The PEM[_ASN1]_{read,write}... functions and macros now take an 6771 additional void * argument, which is just handed through whenever 6772 the password callback is called. 6773 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 6774 6775 New function SSL_CTX_set_default_passwd_cb_userdata. 6776 6777 Compatibility note: As many C implementations push function arguments 6778 onto the stack in reverse order, the new library version is likely to 6779 interoperate with programs that have been compiled with the old 6780 pem_password_cb definition (PEM_whatever takes some data that 6781 happens to be on the stack as its last argument, and the callback 6782 just ignores this garbage); but there is no guarantee whatsoever that 6783 this will work. 6784 6785 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 6786 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 6787 problems not only on Windows, but also on some Unix platforms. 6788 To avoid problematic command lines, these definitions are now in an 6789 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 6790 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 6791 [Bodo Moeller] 6792 6793 *) MIPS III/IV assembler module is reimplemented. 6794 [Andy Polyakov] 6795 6796 *) More DES library cleanups: remove references to srand/rand and 6797 delete an unused file. 6798 [Ulf M�ller] 6799 6800 *) Add support for the the free Netwide assembler (NASM) under Win32, 6801 since not many people have MASM (ml) and it can be hard to obtain. 6802 This is currently experimental but it seems to work OK and pass all 6803 the tests. Check out INSTALL.W32 for info. 6804 [Steve Henson] 6805 6806 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 6807 without temporary keys kept an extra copy of the server key, 6808 and connections with temporary keys did not free everything in case 6809 of an error. 6810 [Bodo Moeller] 6811 6812 *) New function RSA_check_key and new openssl rsa option -check 6813 for verifying the consistency of RSA keys. 6814 [Ulf Moeller, Bodo Moeller] 6815 6816 *) Various changes to make Win32 compile work: 6817 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 6818 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 6819 comparison" warnings. 6820 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 6821 [Steve Henson] 6822 6823 *) Add a debugging option to PKCS#5 v2 key generation function: when 6824 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 6825 derived keys are printed to stderr. 6826 [Steve Henson] 6827 6828 *) Copy the flags in ASN1_STRING_dup(). 6829 [Roman E. Pavlov <pre@mo.msk.ru>] 6830 6831 *) The x509 application mishandled signing requests containing DSA 6832 keys when the signing key was also DSA and the parameters didn't match. 6833 6834 It was supposed to omit the parameters when they matched the signing key: 6835 the verifying software was then supposed to automatically use the CA's 6836 parameters if they were absent from the end user certificate. 6837 6838 Omitting parameters is no longer recommended. The test was also 6839 the wrong way round! This was probably due to unusual behaviour in 6840 EVP_cmp_parameters() which returns 1 if the parameters match. 6841 This meant that parameters were omitted when they *didn't* match and 6842 the certificate was useless. Certificates signed with 'ca' didn't have 6843 this bug. 6844 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 6845 6846 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 6847 The interface is as follows: 6848 Applications can use 6849 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 6850 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 6851 "off" is now the default. 6852 The library internally uses 6853 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 6854 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 6855 to disable memory-checking temporarily. 6856 6857 Some inconsistent states that previously were possible (and were 6858 even the default) are now avoided. 6859 6860 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 6861 with each memory chunk allocated; this is occasionally more helpful 6862 than just having a counter. 6863 6864 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 6865 6866 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 6867 extensions. 6868 [Bodo Moeller] 6869 6870 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 6871 which largely parallels "options", but is for changing API behaviour, 6872 whereas "options" are about protocol behaviour. 6873 Initial "mode" flags are: 6874 6875 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 6876 a single record has been written. 6877 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 6878 retries use the same buffer location. 6879 (But all of the contents must be 6880 copied!) 6881 [Bodo Moeller] 6882 6883 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 6884 worked. 6885 6886 *) Fix problems with no-hmac etc. 6887 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 6888 6889 *) New functions RSA_get_default_method(), RSA_set_method() and 6890 RSA_get_method(). These allows replacement of RSA_METHODs without having 6891 to mess around with the internals of an RSA structure. 6892 [Steve Henson] 6893 6894 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 6895 Also really enable memory leak checks in openssl.c and in some 6896 test programs. 6897 [Chad C. Mulligan, Bodo Moeller] 6898 6899 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 6900 up the length of negative integers. This has now been simplified to just 6901 store the length when it is first determined and use it later, rather 6902 than trying to keep track of where data is copied and updating it to 6903 point to the end. 6904 [Steve Henson, reported by Brien Wheeler 6905 <bwheeler@authentica-security.com>] 6906 6907 *) Add a new function PKCS7_signatureVerify. This allows the verification 6908 of a PKCS#7 signature but with the signing certificate passed to the 6909 function itself. This contrasts with PKCS7_dataVerify which assumes the 6910 certificate is present in the PKCS#7 structure. This isn't always the 6911 case: certificates can be omitted from a PKCS#7 structure and be 6912 distributed by "out of band" means (such as a certificate database). 6913 [Steve Henson] 6914 6915 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 6916 function prototypes in pem.h, also change util/mkdef.pl to add the 6917 necessary function names. 6918 [Steve Henson] 6919 6920 *) mk1mf.pl (used by Windows builds) did not properly read the 6921 options set by Configure in the top level Makefile, and Configure 6922 was not even able to write more than one option correctly. 6923 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 6924 [Bodo Moeller] 6925 6926 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 6927 file to be loaded from a BIO or FILE pointer. The BIO version will 6928 for example allow memory BIOs to contain config info. 6929 [Steve Henson] 6930 6931 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 6932 Whoever hopes to achieve shared-library compatibility across versions 6933 must use this, not the compile-time macro. 6934 (Exercise 0.9.4: Which is the minimum library version required by 6935 such programs?) 6936 Note: All this applies only to multi-threaded programs, others don't 6937 need locks. 6938 [Bodo Moeller] 6939 6940 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 6941 through a BIO pair triggered the default case, i.e. 6942 SSLerr(...,SSL_R_UNKNOWN_STATE). 6943 [Bodo Moeller] 6944 6945 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 6946 can use the SSL library even if none of the specific BIOs is 6947 appropriate. 6948 [Bodo Moeller] 6949 6950 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 6951 for the encoded length. 6952 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 6953 6954 *) Add initial documentation of the X509V3 functions. 6955 [Steve Henson] 6956 6957 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 6958 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 6959 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 6960 secure PKCS#8 private key format with a high iteration count. 6961 [Steve Henson] 6962 6963 *) Fix determination of Perl interpreter: A perl or perl5 6964 _directory_ in $PATH was also accepted as the interpreter. 6965 [Ralf S. Engelschall] 6966 6967 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 6968 wrong with it but it was very old and did things like calling 6969 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 6970 unusual formatting. 6971 [Steve Henson] 6972 6973 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 6974 to use the new extension code. 6975 [Steve Henson] 6976 6977 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 6978 with macros. This should make it easier to change their form, add extra 6979 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 6980 constant. 6981 [Steve Henson] 6982 6983 *) Add to configuration table a new entry that can specify an alternative 6984 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 6985 according to Mark Crispin <MRC@Panda.COM>. 6986 [Bodo Moeller] 6987 6988#if 0 6989 *) DES CBC did not update the IV. Weird. 6990 [Ben Laurie] 6991#else 6992 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 6993 Changing the behaviour of the former might break existing programs -- 6994 where IV updating is needed, des_ncbc_encrypt can be used. 6995#endif 6996 6997 *) When bntest is run from "make test" it drives bc to check its 6998 calculations, as well as internally checking them. If an internal check 6999 fails, it needs to cause bc to give a non-zero result or make test carries 7000 on without noticing the failure. Fixed. 7001 [Ben Laurie] 7002 7003 *) DES library cleanups. 7004 [Ulf M�ller] 7005 7006 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 7007 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 7008 ciphers. NOTE: although the key derivation function has been verified 7009 against some published test vectors it has not been extensively tested 7010 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 7011 of v2.0. 7012 [Steve Henson] 7013 7014 *) Instead of "mkdir -p", which is not fully portable, use new 7015 Perl script "util/mkdir-p.pl". 7016 [Bodo Moeller] 7017 7018 *) Rewrite the way password based encryption (PBE) is handled. It used to 7019 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 7020 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 7021 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 7022 the 'parameter' field of the AlgorithmIdentifier is passed to the 7023 underlying key generation function so it must do its own ASN1 parsing. 7024 This has also changed the EVP_PBE_CipherInit() function which now has a 7025 'parameter' argument instead of literal salt and iteration count values 7026 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 7027 [Steve Henson] 7028 7029 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 7030 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 7031 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 7032 KEY" because this clashed with PKCS#8 unencrypted string. Since this 7033 value was just used as a "magic string" and not used directly its 7034 value doesn't matter. 7035 [Steve Henson] 7036 7037 *) Introduce some semblance of const correctness to BN. Shame C doesn't 7038 support mutable. 7039 [Ben Laurie] 7040 7041 *) "linux-sparc64" configuration (ultrapenguin). 7042 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 7043 "linux-sparc" configuration. 7044 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 7045 7046 *) config now generates no-xxx options for missing ciphers. 7047 [Ulf M�ller] 7048 7049 *) Support the EBCDIC character set (work in progress). 7050 File ebcdic.c not yet included because it has a different license. 7051 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7052 7053 *) Support BS2000/OSD-POSIX. 7054 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7055 7056 *) Make callbacks for key generation use void * instead of char *. 7057 [Ben Laurie] 7058 7059 *) Make S/MIME samples compile (not yet tested). 7060 [Ben Laurie] 7061 7062 *) Additional typesafe stacks. 7063 [Ben Laurie] 7064 7065 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 7066 [Bodo Moeller] 7067 7068 7069 Changes between 0.9.3 and 0.9.3a [29 May 1999] 7070 7071 *) New configuration variant "sco5-gcc". 7072 7073 *) Updated some demos. 7074 [Sean O Riordain, Wade Scholine] 7075 7076 *) Add missing BIO_free at exit of pkcs12 application. 7077 [Wu Zhigang] 7078 7079 *) Fix memory leak in conf.c. 7080 [Steve Henson] 7081 7082 *) Updates for Win32 to assembler version of MD5. 7083 [Steve Henson] 7084 7085 *) Set #! path to perl in apps/der_chop to where we found it 7086 instead of using a fixed path. 7087 [Bodo Moeller] 7088 7089 *) SHA library changes for irix64-mips4-cc. 7090 [Andy Polyakov] 7091 7092 *) Improvements for VMS support. 7093 [Richard Levitte] 7094 7095 7096 Changes between 0.9.2b and 0.9.3 [24 May 1999] 7097 7098 *) Bignum library bug fix. IRIX 6 passes "make test" now! 7099 This also avoids the problems with SC4.2 and unpatched SC5. 7100 [Andy Polyakov <appro@fy.chalmers.se>] 7101 7102 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 7103 These are required because of the typesafe stack would otherwise break 7104 existing code. If old code used a structure member which used to be STACK 7105 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 7106 sk_num or sk_value it would produce an error because the num, data members 7107 are not present in STACK_OF. Now it just produces a warning. sk_set 7108 replaces the old method of assigning a value to sk_value 7109 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 7110 that does this will no longer work (and should use sk_set instead) but 7111 this could be regarded as a "questionable" behaviour anyway. 7112 [Steve Henson] 7113 7114 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 7115 correctly handle encrypted S/MIME data. 7116 [Steve Henson] 7117 7118 *) Change type of various DES function arguments from des_cblock 7119 (which means, in function argument declarations, pointer to char) 7120 to des_cblock * (meaning pointer to array with 8 char elements), 7121 which allows the compiler to do more typechecking; it was like 7122 that back in SSLeay, but with lots of ugly casts. 7123 7124 Introduce new type const_des_cblock. 7125 [Bodo Moeller] 7126 7127 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 7128 problems: find RecipientInfo structure that matches recipient certificate 7129 and initialise the ASN1 structures properly based on passed cipher. 7130 [Steve Henson] 7131 7132 *) Belatedly make the BN tests actually check the results. 7133 [Ben Laurie] 7134 7135 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 7136 to and from BNs: it was completely broken. New compilation option 7137 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 7138 key elements as negative integers. 7139 [Steve Henson] 7140 7141 *) Reorganize and speed up MD5. 7142 [Andy Polyakov <appro@fy.chalmers.se>] 7143 7144 *) VMS support. 7145 [Richard Levitte <richard@levitte.org>] 7146 7147 *) New option -out to asn1parse to allow the parsed structure to be 7148 output to a file. This is most useful when combined with the -strparse 7149 option to examine the output of things like OCTET STRINGS. 7150 [Steve Henson] 7151 7152 *) Make SSL library a little more fool-proof by not requiring any longer 7153 that SSL_set_{accept,connect}_state be called before 7154 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 7155 in many applications because usually everything *appeared* to work as 7156 intended anyway -- now it really works as intended). 7157 [Bodo Moeller] 7158 7159 *) Move openssl.cnf out of lib/. 7160 [Ulf M�ller] 7161 7162 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 7163 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 7164 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 7165 [Ralf S. Engelschall] 7166 7167 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 7168 handle PKCS#7 enveloped data properly. 7169 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 7170 7171 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 7172 copying pointers. The cert_st handling is changed by this in 7173 various ways (and thus what used to be known as ctx->default_cert 7174 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 7175 any longer when s->cert does not give us what we need). 7176 ssl_cert_instantiate becomes obsolete by this change. 7177 As soon as we've got the new code right (possibly it already is?), 7178 we have solved a couple of bugs of the earlier code where s->cert 7179 was used as if it could not have been shared with other SSL structures. 7180 7181 Note that using the SSL API in certain dirty ways now will result 7182 in different behaviour than observed with earlier library versions: 7183 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 7184 does not influence s as it used to. 7185 7186 In order to clean up things more thoroughly, inside SSL_SESSION 7187 we don't use CERT any longer, but a new structure SESS_CERT 7188 that holds per-session data (if available); currently, this is 7189 the peer's certificate chain and, for clients, the server's certificate 7190 and temporary key. CERT holds only those values that can have 7191 meaningful defaults in an SSL_CTX. 7192 [Bodo Moeller] 7193 7194 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 7195 from the internal representation. Various PKCS#7 fixes: remove some 7196 evil casts and set the enc_dig_alg field properly based on the signing 7197 key type. 7198 [Steve Henson] 7199 7200 *) Allow PKCS#12 password to be set from the command line or the 7201 environment. Let 'ca' get its config file name from the environment 7202 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 7203 and 'x509'). 7204 [Steve Henson] 7205 7206 *) Allow certificate policies extension to use an IA5STRING for the 7207 organization field. This is contrary to the PKIX definition but 7208 VeriSign uses it and IE5 only recognises this form. Document 'x509' 7209 extension option. 7210 [Steve Henson] 7211 7212 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 7213 without disallowing inline assembler and the like for non-pedantic builds. 7214 [Ben Laurie] 7215 7216 *) Support Borland C++ builder. 7217 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 7218 7219 *) Support Mingw32. 7220 [Ulf M�ller] 7221 7222 *) SHA-1 cleanups and performance enhancements. 7223 [Andy Polyakov <appro@fy.chalmers.se>] 7224 7225 *) Sparc v8plus assembler for the bignum library. 7226 [Andy Polyakov <appro@fy.chalmers.se>] 7227 7228 *) Accept any -xxx and +xxx compiler options in Configure. 7229 [Ulf M�ller] 7230 7231 *) Update HPUX configuration. 7232 [Anonymous] 7233 7234 *) Add missing sk_<type>_unshift() function to safestack.h 7235 [Ralf S. Engelschall] 7236 7237 *) New function SSL_CTX_use_certificate_chain_file that sets the 7238 "extra_cert"s in addition to the certificate. (This makes sense 7239 only for "PEM" format files, as chains as a whole are not 7240 DER-encoded.) 7241 [Bodo Moeller] 7242 7243 *) Support verify_depth from the SSL API. 7244 x509_vfy.c had what can be considered an off-by-one-error: 7245 Its depth (which was not part of the external interface) 7246 was actually counting the number of certificates in a chain; 7247 now it really counts the depth. 7248 [Bodo Moeller] 7249 7250 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 7251 instead of X509err, which often resulted in confusing error 7252 messages since the error codes are not globally unique 7253 (e.g. an alleged error in ssl3_accept when a certificate 7254 didn't match the private key). 7255 7256 *) New function SSL_CTX_set_session_id_context that allows to set a default 7257 value (so that you don't need SSL_set_session_id_context for each 7258 connection using the SSL_CTX). 7259 [Bodo Moeller] 7260 7261 *) OAEP decoding bug fix. 7262 [Ulf M�ller] 7263 7264 *) Support INSTALL_PREFIX for package builders, as proposed by 7265 David Harris. 7266 [Bodo Moeller] 7267 7268 *) New Configure options "threads" and "no-threads". For systems 7269 where the proper compiler options are known (currently Solaris 7270 and Linux), "threads" is the default. 7271 [Bodo Moeller] 7272 7273 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 7274 [Bodo Moeller] 7275 7276 *) Install various scripts to $(OPENSSLDIR)/misc, not to 7277 $(INSTALLTOP)/bin -- they shouldn't clutter directories 7278 such as /usr/local/bin. 7279 [Bodo Moeller] 7280 7281 *) "make linux-shared" to build shared libraries. 7282 [Niels Poppe <niels@netbox.org>] 7283 7284 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 7285 [Ulf M�ller] 7286 7287 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 7288 extension adding in x509 utility. 7289 [Steve Henson] 7290 7291 *) Remove NOPROTO sections and error code comments. 7292 [Ulf M�ller] 7293 7294 *) Partial rewrite of the DEF file generator to now parse the ANSI 7295 prototypes. 7296 [Steve Henson] 7297 7298 *) New Configure options --prefix=DIR and --openssldir=DIR. 7299 [Ulf M�ller] 7300 7301 *) Complete rewrite of the error code script(s). It is all now handled 7302 by one script at the top level which handles error code gathering, 7303 header rewriting and C source file generation. It should be much better 7304 than the old method: it now uses a modified version of Ulf's parser to 7305 read the ANSI prototypes in all header files (thus the old K&R definitions 7306 aren't needed for error creation any more) and do a better job of 7307 translating function codes into names. The old 'ASN1 error code imbedded 7308 in a comment' is no longer necessary and it doesn't use .err files which 7309 have now been deleted. Also the error code call doesn't have to appear all 7310 on one line (which resulted in some large lines...). 7311 [Steve Henson] 7312 7313 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 7314 [Bodo Moeller] 7315 7316 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 7317 0 (which usually indicates a closed connection), but continue reading. 7318 [Bodo Moeller] 7319 7320 *) Fix some race conditions. 7321 [Bodo Moeller] 7322 7323 *) Add support for CRL distribution points extension. Add Certificate 7324 Policies and CRL distribution points documentation. 7325 [Steve Henson] 7326 7327 *) Move the autogenerated header file parts to crypto/opensslconf.h. 7328 [Ulf M�ller] 7329 7330 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 7331 8 of keying material. Merlin has also confirmed interop with this fix 7332 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 7333 [Merlin Hughes <merlin@baltimore.ie>] 7334 7335 *) Fix lots of warnings. 7336 [Richard Levitte <levitte@stacken.kth.se>] 7337 7338 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 7339 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 7340 [Richard Levitte <levitte@stacken.kth.se>] 7341 7342 *) Fix problems with sizeof(long) == 8. 7343 [Andy Polyakov <appro@fy.chalmers.se>] 7344 7345 *) Change functions to ANSI C. 7346 [Ulf M�ller] 7347 7348 *) Fix typos in error codes. 7349 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 7350 7351 *) Remove defunct assembler files from Configure. 7352 [Ulf M�ller] 7353 7354 *) SPARC v8 assembler BIGNUM implementation. 7355 [Andy Polyakov <appro@fy.chalmers.se>] 7356 7357 *) Support for Certificate Policies extension: both print and set. 7358 Various additions to support the r2i method this uses. 7359 [Steve Henson] 7360 7361 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 7362 return a const string when you are expecting an allocated buffer. 7363 [Ben Laurie] 7364 7365 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 7366 types DirectoryString and DisplayText. 7367 [Steve Henson] 7368 7369 *) Add code to allow r2i extensions to access the configuration database, 7370 add an LHASH database driver and add several ctx helper functions. 7371 [Steve Henson] 7372 7373 *) Fix an evil bug in bn_expand2() which caused various BN functions to 7374 fail when they extended the size of a BIGNUM. 7375 [Steve Henson] 7376 7377 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 7378 support typesafe stack. 7379 [Steve Henson] 7380 7381 *) Fix typo in SSL_[gs]et_options(). 7382 [Nils Frostberg <nils@medcom.se>] 7383 7384 *) Delete various functions and files that belonged to the (now obsolete) 7385 old X509V3 handling code. 7386 [Steve Henson] 7387 7388 *) New Configure option "rsaref". 7389 [Ulf M�ller] 7390 7391 *) Don't auto-generate pem.h. 7392 [Bodo Moeller] 7393 7394 *) Introduce type-safe ASN.1 SETs. 7395 [Ben Laurie] 7396 7397 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 7398 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 7399 7400 *) Introduce type-safe STACKs. This will almost certainly break lots of code 7401 that links with OpenSSL (well at least cause lots of warnings), but fear 7402 not: the conversion is trivial, and it eliminates loads of evil casts. A 7403 few STACKed things have been converted already. Feel free to convert more. 7404 In the fullness of time, I'll do away with the STACK type altogether. 7405 [Ben Laurie] 7406 7407 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 7408 specified in <certfile> by updating the entry in the index.txt file. 7409 This way one no longer has to edit the index.txt file manually for 7410 revoking a certificate. The -revoke option does the gory details now. 7411 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 7412 7413 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 7414 `-text' option at all and this way the `-noout -text' combination was 7415 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 7416 [Ralf S. Engelschall] 7417 7418 *) Make sure a corresponding plain text error message exists for the 7419 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 7420 verify callback function determined that a certificate was revoked. 7421 [Ralf S. Engelschall] 7422 7423 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 7424 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 7425 all available cipers including rc5, which was forgotten until now. 7426 In order to let the testing shell script know which algorithms 7427 are available, a new (up to now undocumented) command 7428 "openssl list-cipher-commands" is used. 7429 [Bodo Moeller] 7430 7431 *) Bugfix: s_client occasionally would sleep in select() when 7432 it should have checked SSL_pending() first. 7433 [Bodo Moeller] 7434 7435 *) New functions DSA_do_sign and DSA_do_verify to provide access to 7436 the raw DSA values prior to ASN.1 encoding. 7437 [Ulf M�ller] 7438 7439 *) Tweaks to Configure 7440 [Niels Poppe <niels@netbox.org>] 7441 7442 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 7443 yet... 7444 [Steve Henson] 7445 7446 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 7447 [Ulf M�ller] 7448 7449 *) New config option to avoid instructions that are illegal on the 80386. 7450 The default code is faster, but requires at least a 486. 7451 [Ulf M�ller] 7452 7453 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 7454 SSL2_SERVER_VERSION (not used at all) macros, which are now the 7455 same as SSL2_VERSION anyway. 7456 [Bodo Moeller] 7457 7458 *) New "-showcerts" option for s_client. 7459 [Bodo Moeller] 7460 7461 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 7462 application. Various cleanups and fixes. 7463 [Steve Henson] 7464 7465 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 7466 modify error routines to work internally. Add error codes and PBE init 7467 to library startup routines. 7468 [Steve Henson] 7469 7470 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 7471 packing functions to asn1 and evp. Changed function names and error 7472 codes along the way. 7473 [Steve Henson] 7474 7475 *) PKCS12 integration: and so it begins... First of several patches to 7476 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 7477 objects to objects.h 7478 [Steve Henson] 7479 7480 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 7481 and display support for Thawte strong extranet extension. 7482 [Steve Henson] 7483 7484 *) Add LinuxPPC support. 7485 [Jeff Dubrule <igor@pobox.org>] 7486 7487 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 7488 bn_div_words in alpha.s. 7489 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 7490 7491 *) Make sure the RSA OAEP test is skipped under -DRSAref because 7492 OAEP isn't supported when OpenSSL is built with RSAref. 7493 [Ulf Moeller <ulf@fitug.de>] 7494 7495 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 7496 so they no longer are missing under -DNOPROTO. 7497 [Soren S. Jorvang <soren@t.dk>] 7498 7499 7500 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 7501 7502 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 7503 doesn't work when the session is reused. Coming soon! 7504 [Ben Laurie] 7505 7506 *) Fix a security hole, that allows sessions to be reused in the wrong 7507 context thus bypassing client cert protection! All software that uses 7508 client certs and session caches in multiple contexts NEEDS PATCHING to 7509 allow session reuse! A fuller solution is in the works. 7510 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 7511 7512 *) Some more source tree cleanups (removed obsolete files 7513 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 7514 permission on "config" script to be executable) and a fix for the INSTALL 7515 document. 7516 [Ulf Moeller <ulf@fitug.de>] 7517 7518 *) Remove some legacy and erroneous uses of malloc, free instead of 7519 Malloc, Free. 7520 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 7521 7522 *) Make rsa_oaep_test return non-zero on error. 7523 [Ulf Moeller <ulf@fitug.de>] 7524 7525 *) Add support for native Solaris shared libraries. Configure 7526 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 7527 if someone would make that last step automatic. 7528 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 7529 7530 *) ctx_size was not built with the right compiler during "make links". Fixed. 7531 [Ben Laurie] 7532 7533 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 7534 except NULL ciphers". This means the default cipher list will no longer 7535 enable NULL ciphers. They need to be specifically enabled e.g. with 7536 the string "DEFAULT:eNULL". 7537 [Steve Henson] 7538 7539 *) Fix to RSA private encryption routines: if p < q then it would 7540 occasionally produce an invalid result. This will only happen with 7541 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 7542 [Steve Henson] 7543 7544 *) Be less restrictive and allow also `perl util/perlpath.pl 7545 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 7546 because this way one can also use an interpreter named `perl5' (which is 7547 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 7548 installed as `perl'). 7549 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7550 7551 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 7552 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7553 7554 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 7555 advapi32.lib to Win32 build and change the pem test comparision 7556 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 7557 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 7558 and crypto/des/ede_cbcm_enc.c. 7559 [Steve Henson] 7560 7561 *) DES quad checksum was broken on big-endian architectures. Fixed. 7562 [Ben Laurie] 7563 7564 *) Comment out two functions in bio.h that aren't implemented. Fix up the 7565 Win32 test batch file so it (might) work again. The Win32 test batch file 7566 is horrible: I feel ill.... 7567 [Steve Henson] 7568 7569 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 7570 in e_os.h. Audit of header files to check ANSI and non ANSI 7571 sections: 10 functions were absent from non ANSI section and not exported 7572 from Windows DLLs. Fixed up libeay.num for new functions. 7573 [Steve Henson] 7574 7575 *) Make `openssl version' output lines consistent. 7576 [Ralf S. Engelschall] 7577 7578 *) Fix Win32 symbol export lists for BIO functions: Added 7579 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 7580 to ms/libeay{16,32}.def. 7581 [Ralf S. Engelschall] 7582 7583 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 7584 fine under Unix and passes some trivial tests I've now added. But the 7585 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 7586 added to make sure no one expects that this stuff really works in the 7587 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 7588 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 7589 openssl_bio.xs. 7590 [Ralf S. Engelschall] 7591 7592 *) Fix the generation of two part addresses in perl. 7593 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 7594 7595 *) Add config entry for Linux on MIPS. 7596 [John Tobey <jtobey@channel1.com>] 7597 7598 *) Make links whenever Configure is run, unless we are on Windoze. 7599 [Ben Laurie] 7600 7601 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 7602 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 7603 in CRLs. 7604 [Steve Henson] 7605 7606 *) Add a useful kludge to allow package maintainers to specify compiler and 7607 other platforms details on the command line without having to patch the 7608 Configure script everytime: One now can use ``perl Configure 7609 <id>:<details>'', i.e. platform ids are allowed to have details appended 7610 to them (seperated by colons). This is treated as there would be a static 7611 pre-configured entry in Configure's %table under key <id> with value 7612 <details> and ``perl Configure <id>'' is called. So, when you want to 7613 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 7614 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 7615 now, which overrides the FreeBSD-elf entry on-the-fly. 7616 [Ralf S. Engelschall] 7617 7618 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 7619 [Ben Laurie] 7620 7621 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 7622 on the `perl Configure ...' command line. This way one can compile 7623 OpenSSL libraries with Position Independent Code (PIC) which is needed 7624 for linking it into DSOs. 7625 [Ralf S. Engelschall] 7626 7627 *) Remarkably, export ciphers were totally broken and no-one had noticed! 7628 Fixed. 7629 [Ben Laurie] 7630 7631 *) Cleaned up the LICENSE document: The official contact for any license 7632 questions now is the OpenSSL core team under openssl-core@openssl.org. 7633 And add a paragraph about the dual-license situation to make sure people 7634 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 7635 to the OpenSSL toolkit. 7636 [Ralf S. Engelschall] 7637 7638 *) General source tree makefile cleanups: Made `making xxx in yyy...' 7639 display consistent in the source tree and replaced `/bin/rm' by `rm'. 7640 Additonally cleaned up the `make links' target: Remove unnecessary 7641 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 7642 to speed processing and no longer clutter the display with confusing 7643 stuff. Instead only the actually done links are displayed. 7644 [Ralf S. Engelschall] 7645 7646 *) Permit null encryption ciphersuites, used for authentication only. It used 7647 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 7648 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 7649 encryption. 7650 [Ben Laurie] 7651 7652 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 7653 signed attributes when verifying signatures (this would break them), 7654 the detached data encoding was wrong and public keys obtained using 7655 X509_get_pubkey() weren't freed. 7656 [Steve Henson] 7657 7658 *) Add text documentation for the BUFFER functions. Also added a work around 7659 to a Win95 console bug. This was triggered by the password read stuff: the 7660 last character typed gets carried over to the next fread(). If you were 7661 generating a new cert request using 'req' for example then the last 7662 character of the passphrase would be CR which would then enter the first 7663 field as blank. 7664 [Steve Henson] 7665 7666 *) Added the new `Includes OpenSSL Cryptography Software' button as 7667 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 7668 button and can be used by applications based on OpenSSL to show the 7669 relationship to the OpenSSL project. 7670 [Ralf S. Engelschall] 7671 7672 *) Remove confusing variables in function signatures in files 7673 ssl/ssl_lib.c and ssl/ssl.h. 7674 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7675 7676 *) Don't install bss_file.c under PREFIX/include/ 7677 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7678 7679 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 7680 functions that return function pointers and has support for NT specific 7681 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 7682 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 7683 unsigned to signed types: this was killing the Win32 compile. 7684 [Steve Henson] 7685 7686 *) Add new certificate file to stack functions, 7687 SSL_add_dir_cert_subjects_to_stack() and 7688 SSL_add_file_cert_subjects_to_stack(). These largely supplant 7689 SSL_load_client_CA_file(), and can be used to add multiple certs easily 7690 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 7691 This means that Apache-SSL and similar packages don't have to mess around 7692 to add as many CAs as they want to the preferred list. 7693 [Ben Laurie] 7694 7695 *) Experiment with doxygen documentation. Currently only partially applied to 7696 ssl/ssl_lib.c. 7697 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 7698 openssl.doxy as the configuration file. 7699 [Ben Laurie] 7700 7701 *) Get rid of remaining C++-style comments which strict C compilers hate. 7702 [Ralf S. Engelschall, pointed out by Carlos Amengual] 7703 7704 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 7705 compiled in by default: it has problems with large keys. 7706 [Steve Henson] 7707 7708 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 7709 DH private keys and/or callback functions which directly correspond to 7710 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 7711 is needed for applications which have to configure certificates on a 7712 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 7713 (e.g. s_server). 7714 For the RSA certificate situation is makes no difference, but 7715 for the DSA certificate situation this fixes the "no shared cipher" 7716 problem where the OpenSSL cipher selection procedure failed because the 7717 temporary keys were not overtaken from the context and the API provided 7718 no way to reconfigure them. 7719 The new functions now let applications reconfigure the stuff and they 7720 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 7721 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 7722 non-public-API function ssl_cert_instantiate() is used as a helper 7723 function and also to reduce code redundancy inside ssl_rsa.c. 7724 [Ralf S. Engelschall] 7725 7726 *) Move s_server -dcert and -dkey options out of the undocumented feature 7727 area because they are useful for the DSA situation and should be 7728 recognized by the users. 7729 [Ralf S. Engelschall] 7730 7731 *) Fix the cipher decision scheme for export ciphers: the export bits are 7732 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 7733 SSL_EXP_MASK. So, the original variable has to be used instead of the 7734 already masked variable. 7735 [Richard Levitte <levitte@stacken.kth.se>] 7736 7737 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 7738 [Richard Levitte <levitte@stacken.kth.se>] 7739 7740 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 7741 from `int' to `unsigned int' because it's a length and initialized by 7742 EVP_DigestFinal() which expects an `unsigned int *'. 7743 [Richard Levitte <levitte@stacken.kth.se>] 7744 7745 *) Don't hard-code path to Perl interpreter on shebang line of Configure 7746 script. Instead use the usual Shell->Perl transition trick. 7747 [Ralf S. Engelschall] 7748 7749 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 7750 (in addition to RSA certificates) to match the behaviour of `openssl dsa 7751 -noout -modulus' as it's already the case for `openssl rsa -noout 7752 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 7753 currently the public key is printed (a decision which was already done by 7754 `openssl dsa -modulus' in the past) which serves a similar purpose. 7755 Additionally the NO_RSA no longer completely removes the whole -modulus 7756 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 7757 now, too. 7758 [Ralf S. Engelschall] 7759 7760 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 7761 BIO. See the source (crypto/evp/bio_ok.c) for more info. 7762 [Arne Ansper <arne@ats.cyber.ee>] 7763 7764 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 7765 to be added. Now both 'req' and 'ca' can use new objects defined in the 7766 config file. 7767 [Steve Henson] 7768 7769 *) Add cool BIO that does syslog (or event log on NT). 7770 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 7771 7772 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 7773 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 7774 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 7775 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 7776 [Ben Laurie] 7777 7778 *) Add preliminary config info for new extension code. 7779 [Steve Henson] 7780 7781 *) Make RSA_NO_PADDING really use no padding. 7782 [Ulf Moeller <ulf@fitug.de>] 7783 7784 *) Generate errors when private/public key check is done. 7785 [Ben Laurie] 7786 7787 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 7788 for some CRL extensions and new objects added. 7789 [Steve Henson] 7790 7791 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 7792 key usage extension and fuller support for authority key id. 7793 [Steve Henson] 7794 7795 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 7796 padding method for RSA, which is recommended for new applications in PKCS 7797 #1 v2.0 (RFC 2437, October 1998). 7798 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 7799 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 7800 against Bleichbacher's attack on RSA. 7801 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 7802 Ben Laurie] 7803 7804 *) Updates to the new SSL compression code 7805 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7806 7807 *) Fix so that the version number in the master secret, when passed 7808 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 7809 (because the server will not accept higher), that the version number 7810 is 0x03,0x01, not 0x03,0x00 7811 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7812 7813 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 7814 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 7815 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 7816 [Steve Henson] 7817 7818 *) Support for RAW extensions where an arbitrary extension can be 7819 created by including its DER encoding. See apps/openssl.cnf for 7820 an example. 7821 [Steve Henson] 7822 7823 *) Make sure latest Perl versions don't interpret some generated C array 7824 code as Perl array code in the crypto/err/err_genc.pl script. 7825 [Lars Weber <3weber@informatik.uni-hamburg.de>] 7826 7827 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 7828 not many people have the assembler. Various Win32 compilation fixes and 7829 update to the INSTALL.W32 file with (hopefully) more accurate Win32 7830 build instructions. 7831 [Steve Henson] 7832 7833 *) Modify configure script 'Configure' to automatically create crypto/date.h 7834 file under Win32 and also build pem.h from pem.org. New script 7835 util/mkfiles.pl to create the MINFO file on environments that can't do a 7836 'make files': perl util/mkfiles.pl >MINFO should work. 7837 [Steve Henson] 7838 7839 *) Major rework of DES function declarations, in the pursuit of correctness 7840 and purity. As a result, many evil casts evaporated, and some weirdness, 7841 too. You may find this causes warnings in your code. Zapping your evil 7842 casts will probably fix them. Mostly. 7843 [Ben Laurie] 7844 7845 *) Fix for a typo in asn1.h. Bug fix to object creation script 7846 obj_dat.pl. It considered a zero in an object definition to mean 7847 "end of object": none of the objects in objects.h have any zeros 7848 so it wasn't spotted. 7849 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 7850 7851 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 7852 Masking (CBCM). In the absence of test vectors, the best I have been able 7853 to do is check that the decrypt undoes the encrypt, so far. Send me test 7854 vectors if you have them. 7855 [Ben Laurie] 7856 7857 *) Correct calculation of key length for export ciphers (too much space was 7858 allocated for null ciphers). This has not been tested! 7859 [Ben Laurie] 7860 7861 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 7862 message is now correct (it understands "crypto" and "ssl" on its 7863 command line). There is also now an "update" option. This will update 7864 the util/ssleay.num and util/libeay.num files with any new functions. 7865 If you do a: 7866 perl util/mkdef.pl crypto ssl update 7867 it will update them. 7868 [Steve Henson] 7869 7870 *) Overhauled the Perl interface (perl/*): 7871 - ported BN stuff to OpenSSL's different BN library 7872 - made the perl/ source tree CVS-aware 7873 - renamed the package from SSLeay to OpenSSL (the files still contain 7874 their history because I've copied them in the repository) 7875 - removed obsolete files (the test scripts will be replaced 7876 by better Test::Harness variants in the future) 7877 [Ralf S. Engelschall] 7878 7879 *) First cut for a very conservative source tree cleanup: 7880 1. merge various obsolete readme texts into doc/ssleay.txt 7881 where we collect the old documents and readme texts. 7882 2. remove the first part of files where I'm already sure that we no 7883 longer need them because of three reasons: either they are just temporary 7884 files which were left by Eric or they are preserved original files where 7885 I've verified that the diff is also available in the CVS via "cvs diff 7886 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 7887 the crypto/md/ stuff). 7888 [Ralf S. Engelschall] 7889 7890 *) More extension code. Incomplete support for subject and issuer alt 7891 name, issuer and authority key id. Change the i2v function parameters 7892 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 7893 what that's for :-) Fix to ASN1 macro which messed up 7894 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 7895 [Steve Henson] 7896 7897 *) Preliminary support for ENUMERATED type. This is largely copied from the 7898 INTEGER code. 7899 [Steve Henson] 7900 7901 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 7902 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7903 7904 *) Make sure `make rehash' target really finds the `openssl' program. 7905 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7906 7907 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 7908 like to hear about it if this slows down other processors. 7909 [Ben Laurie] 7910 7911 *) Add CygWin32 platform information to Configure script. 7912 [Alan Batie <batie@aahz.jf.intel.com>] 7913 7914 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 7915 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 7916 7917 *) New program nseq to manipulate netscape certificate sequences 7918 [Steve Henson] 7919 7920 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 7921 few typos. 7922 [Steve Henson] 7923 7924 *) Fixes to BN code. Previously the default was to define BN_RECURSION 7925 but the BN code had some problems that would cause failures when 7926 doing certificate verification and some other functions. 7927 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7928 7929 *) Add ASN1 and PEM code to support netscape certificate sequences. 7930 [Steve Henson] 7931 7932 *) Add ASN1 and PEM code to support netscape certificate sequences. 7933 [Steve Henson] 7934 7935 *) Add several PKIX and private extended key usage OIDs. 7936 [Steve Henson] 7937 7938 *) Modify the 'ca' program to handle the new extension code. Modify 7939 openssl.cnf for new extension format, add comments. 7940 [Steve Henson] 7941 7942 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 7943 and add a sample to openssl.cnf so req -x509 now adds appropriate 7944 CA extensions. 7945 [Steve Henson] 7946 7947 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 7948 error code, add initial support to X509_print() and x509 application. 7949 [Steve Henson] 7950 7951 *) Takes a deep breath and start addding X509 V3 extension support code. Add 7952 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 7953 stuff is currently isolated and isn't even compiled yet. 7954 [Steve Henson] 7955 7956 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 7957 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 7958 Removed the versions check from X509 routines when loading extensions: 7959 this allows certain broken certificates that don't set the version 7960 properly to be processed. 7961 [Steve Henson] 7962 7963 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 7964 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 7965 can still be regenerated with "make depend". 7966 [Ben Laurie] 7967 7968 *) Spelling mistake in C version of CAST-128. 7969 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 7970 7971 *) Changes to the error generation code. The perl script err-code.pl 7972 now reads in the old error codes and retains the old numbers, only 7973 adding new ones if necessary. It also only changes the .err files if new 7974 codes are added. The makefiles have been modified to only insert errors 7975 when needed (to avoid needlessly modifying header files). This is done 7976 by only inserting errors if the .err file is newer than the auto generated 7977 C file. To rebuild all the error codes from scratch (the old behaviour) 7978 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 7979 or delete all the .err files. 7980 [Steve Henson] 7981 7982 *) CAST-128 was incorrectly implemented for short keys. The C version has 7983 been fixed, but is untested. The assembler versions are also fixed, but 7984 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 7985 to regenerate it if needed. 7986 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 7987 Hagino <itojun@kame.net>] 7988 7989 *) File was opened incorrectly in randfile.c. 7990 [Ulf M�ller <ulf@fitug.de>] 7991 7992 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 7993 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 7994 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 7995 al: it's just almost always a UTCTime. Note this patch adds new error 7996 codes so do a "make errors" if there are problems. 7997 [Steve Henson] 7998 7999 *) Correct Linux 1 recognition in config. 8000 [Ulf M�ller <ulf@fitug.de>] 8001 8002 *) Remove pointless MD5 hash when using DSA keys in ca. 8003 [Anonymous <nobody@replay.com>] 8004 8005 *) Generate an error if given an empty string as a cert directory. Also 8006 generate an error if handed NULL (previously returned 0 to indicate an 8007 error, but didn't set one). 8008 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 8009 8010 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 8011 [Ben Laurie] 8012 8013 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 8014 parameters. This was causing a warning which killed off the Win32 compile. 8015 [Steve Henson] 8016 8017 *) Remove C++ style comments from crypto/bn/bn_local.h. 8018 [Neil Costigan <neil.costigan@celocom.com>] 8019 8020 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 8021 based on a text string, looking up short and long names and finally 8022 "dot" format. The "dot" format stuff didn't work. Added new function 8023 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 8024 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 8025 OID is not part of the table. 8026 [Steve Henson] 8027 8028 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 8029 X509_LOOKUP_by_alias(). 8030 [Ben Laurie] 8031 8032 *) Sort openssl functions by name. 8033 [Ben Laurie] 8034 8035 *) Get the gendsa program working (hopefully) and add it to app list. Remove 8036 encryption from sample DSA keys (in case anyone is interested the password 8037 was "1234"). 8038 [Steve Henson] 8039 8040 *) Make _all_ *_free functions accept a NULL pointer. 8041 [Frans Heymans <fheymans@isaserver.be>] 8042 8043 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 8044 NULL pointers. 8045 [Anonymous <nobody@replay.com>] 8046 8047 *) s_server should send the CAfile as acceptable CAs, not its own cert. 8048 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8049 8050 *) Don't blow it for numeric -newkey arguments to apps/req. 8051 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8052 8053 *) Temp key "for export" tests were wrong in s3_srvr.c. 8054 [Anonymous <nobody@replay.com>] 8055 8056 *) Add prototype for temp key callback functions 8057 SSL_CTX_set_tmp_{rsa,dh}_callback(). 8058 [Ben Laurie] 8059 8060 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 8061 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 8062 [Steve Henson] 8063 8064 *) X509_name_add_entry() freed the wrong thing after an error. 8065 [Arne Ansper <arne@ats.cyber.ee>] 8066 8067 *) rsa_eay.c would attempt to free a NULL context. 8068 [Arne Ansper <arne@ats.cyber.ee>] 8069 8070 *) BIO_s_socket() had a broken should_retry() on Windoze. 8071 [Arne Ansper <arne@ats.cyber.ee>] 8072 8073 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 8074 [Arne Ansper <arne@ats.cyber.ee>] 8075 8076 *) Make sure the already existing X509_STORE->depth variable is initialized 8077 in X509_STORE_new(), but document the fact that this variable is still 8078 unused in the certificate verification process. 8079 [Ralf S. Engelschall] 8080 8081 *) Fix the various library and apps files to free up pkeys obtained from 8082 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 8083 [Steve Henson] 8084 8085 *) Fix reference counting in X509_PUBKEY_get(). This makes 8086 demos/maurice/example2.c work, amongst others, probably. 8087 [Steve Henson and Ben Laurie] 8088 8089 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 8090 `openssl' and second, the shortcut symlinks for the `openssl <command>' 8091 are no longer created. This way we have a single and consistent command 8092 line interface `openssl <command>', similar to `cvs <command>'. 8093 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 8094 8095 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 8096 BIT STRING wrapper always have zero unused bits. 8097 [Steve Henson] 8098 8099 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 8100 [Steve Henson] 8101 8102 *) Make the top-level INSTALL documentation easier to understand. 8103 [Paul Sutton] 8104 8105 *) Makefiles updated to exit if an error occurs in a sub-directory 8106 make (including if user presses ^C) [Paul Sutton] 8107 8108 *) Make Montgomery context stuff explicit in RSA data structure. 8109 [Ben Laurie] 8110 8111 *) Fix build order of pem and err to allow for generated pem.h. 8112 [Ben Laurie] 8113 8114 *) Fix renumbering bug in X509_NAME_delete_entry(). 8115 [Ben Laurie] 8116 8117 *) Enhanced the err-ins.pl script so it makes the error library number 8118 global and can add a library name. This is needed for external ASN1 and 8119 other error libraries. 8120 [Steve Henson] 8121 8122 *) Fixed sk_insert which never worked properly. 8123 [Steve Henson] 8124 8125 *) Fix ASN1 macros so they can handle indefinite length construted 8126 EXPLICIT tags. Some non standard certificates use these: they can now 8127 be read in. 8128 [Steve Henson] 8129 8130 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 8131 into a single doc/ssleay.txt bundle. This way the information is still 8132 preserved but no longer messes up this directory. Now it's new room for 8133 the new set of documenation files. 8134 [Ralf S. Engelschall] 8135 8136 *) SETs were incorrectly DER encoded. This was a major pain, because they 8137 shared code with SEQUENCEs, which aren't coded the same. This means that 8138 almost everything to do with SETs or SEQUENCEs has either changed name or 8139 number of arguments. 8140 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 8141 8142 *) Fix test data to work with the above. 8143 [Ben Laurie] 8144 8145 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 8146 was already fixed by Eric for 0.9.1 it seems. 8147 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 8148 8149 *) Autodetect FreeBSD3. 8150 [Ben Laurie] 8151 8152 *) Fix various bugs in Configure. This affects the following platforms: 8153 nextstep 8154 ncr-scde 8155 unixware-2.0 8156 unixware-2.0-pentium 8157 sco5-cc. 8158 [Ben Laurie] 8159 8160 *) Eliminate generated files from CVS. Reorder tests to regenerate files 8161 before they are needed. 8162 [Ben Laurie] 8163 8164 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 8165 [Ben Laurie] 8166 8167 8168 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 8169 8170 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 8171 changed SSLeay to OpenSSL in version strings. 8172 [Ralf S. Engelschall] 8173 8174 *) Some fixups to the top-level documents. 8175 [Paul Sutton] 8176 8177 *) Fixed the nasty bug where rsaref.h was not found under compile-time 8178 because the symlink to include/ was missing. 8179 [Ralf S. Engelschall] 8180 8181 *) Incorporated the popular no-RSA/DSA-only patches 8182 which allow to compile a RSA-free SSLeay. 8183 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 8184 8185 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 8186 when "ssleay" is still not found. 8187 [Ralf S. Engelschall] 8188 8189 *) Added more platforms to Configure: Cray T3E, HPUX 11, 8190 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 8191 8192 *) Updated the README file. 8193 [Ralf S. Engelschall] 8194 8195 *) Added various .cvsignore files in the CVS repository subdirs 8196 to make a "cvs update" really silent. 8197 [Ralf S. Engelschall] 8198 8199 *) Recompiled the error-definition header files and added 8200 missing symbols to the Win32 linker tables. 8201 [Ralf S. Engelschall] 8202 8203 *) Cleaned up the top-level documents; 8204 o new files: CHANGES and LICENSE 8205 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 8206 o merged COPYRIGHT into LICENSE 8207 o removed obsolete TODO file 8208 o renamed MICROSOFT to INSTALL.W32 8209 [Ralf S. Engelschall] 8210 8211 *) Removed dummy files from the 0.9.1b source tree: 8212 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 8213 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 8214 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 8215 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 8216 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 8217 [Ralf S. Engelschall] 8218 8219 *) Added various platform portability fixes. 8220 [Mark J. Cox] 8221 8222 *) The Genesis of the OpenSSL rpject: 8223 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 8224 Young and Tim J. Hudson created while they were working for C2Net until 8225 summer 1998. 8226 [The OpenSSL Project] 8227 8228 8229 Changes between 0.9.0b and 0.9.1b [not released] 8230 8231 *) Updated a few CA certificates under certs/ 8232 [Eric A. Young] 8233 8234 *) Changed some BIGNUM api stuff. 8235 [Eric A. Young] 8236 8237 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 8238 DGUX x86, Linux Alpha, etc. 8239 [Eric A. Young] 8240 8241 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 8242 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 8243 available). 8244 [Eric A. Young] 8245 8246 *) Add -strparse option to asn1pars program which parses nested 8247 binary structures 8248 [Dr Stephen Henson <shenson@bigfoot.com>] 8249 8250 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 8251 [Eric A. Young] 8252 8253 *) DSA fix for "ca" program. 8254 [Eric A. Young] 8255 8256 *) Added "-genkey" option to "dsaparam" program. 8257 [Eric A. Young] 8258 8259 *) Added RIPE MD160 (rmd160) message digest. 8260 [Eric A. Young] 8261 8262 *) Added -a (all) option to "ssleay version" command. 8263 [Eric A. Young] 8264 8265 *) Added PLATFORM define which is the id given to Configure. 8266 [Eric A. Young] 8267 8268 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 8269 [Eric A. Young] 8270 8271 *) Extended the ASN.1 parser routines. 8272 [Eric A. Young] 8273 8274 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 8275 [Eric A. Young] 8276 8277 *) Added a BN_CTX to the BN library. 8278 [Eric A. Young] 8279 8280 *) Fixed the weak key values in DES library 8281 [Eric A. Young] 8282 8283 *) Changed API in EVP library for cipher aliases. 8284 [Eric A. Young] 8285 8286 *) Added support for RC2/64bit cipher. 8287 [Eric A. Young] 8288 8289 *) Converted the lhash library to the crypto/mem.c functions. 8290 [Eric A. Young] 8291 8292 *) Added more recognized ASN.1 object ids. 8293 [Eric A. Young] 8294 8295 *) Added more RSA padding checks for SSL/TLS. 8296 [Eric A. Young] 8297 8298 *) Added BIO proxy/filter functionality. 8299 [Eric A. Young] 8300 8301 *) Added extra_certs to SSL_CTX which can be used 8302 send extra CA certificates to the client in the CA cert chain sending 8303 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 8304 [Eric A. Young] 8305 8306 *) Now Fortezza is denied in the authentication phase because 8307 this is key exchange mechanism is not supported by SSLeay at all. 8308 [Eric A. Young] 8309 8310 *) Additional PKCS1 checks. 8311 [Eric A. Young] 8312 8313 *) Support the string "TLSv1" for all TLS v1 ciphers. 8314 [Eric A. Young] 8315 8316 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 8317 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 8318 [Eric A. Young] 8319 8320 *) Fixed a few memory leaks. 8321 [Eric A. Young] 8322 8323 *) Fixed various code and comment typos. 8324 [Eric A. Young] 8325 8326 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 8327 bytes sent in the client random. 8328 [Edward Bishop <ebishop@spyglass.com>] 8329 8330